RTEMS User Manual (5.1).¶
Copyrights and License
This document is available under the Creative Commons Attribution-ShareAlike 4.0 International Public License.
The authors have used their best efforts in preparing this material. These efforts include the development, research, and testing of the theories and programs to determine their effectiveness. No warranty of any kind, expressed or implied, with regard to the software or the material contained in this document is provided. No liability arising out of the application or use of any product described in this document is assumed. The authors reserve the right to revise this material and to make changes from time to time in the content hereof without obligation to notify anyone of such revision or changes.
The RTEMS Project is hosted at https://www.rtems.org. Any inquiries concerning RTEMS, its related support components, or its documentation should be directed to the RTEMS Project community.
RTEMS Online Resources
Home | |
Documentation | |
Mailing Lists | |
Bug Reporting | |
Git Repositories | |
Developers |
1. Introduction¶
1.1. Overview¶
You are someone looking for a real-time operating system. This document
presents the basic features of RTEMS, so that you can decide if it is worth to look at,
gives you a quick start to install all the tools necessary to work with RTEMS, and
helps you to build an example application on top of RTEMS.
1.2. Features¶
The Real-Time Executive for Multiprocessor Systems (RTEMS) is a multi-threaded, single address-space, real-time operating system with no kernel-space/user-space separation. It is capable to operate in an SMP configuration providing a state of the art feature set.
RTEMS is licensed under a modified GPL 2.0 or later license with an exception for static linking 1. It exposes no license requirements on application code. The third-party software used and distributed by RTEMS which may be linked to the application is licensed under permissive open source licenses. Everything necessary to build RTEMS applications is available as open source software. This makes you completely vendor independent.
RTEMS provides the following basic feature set:
-
Programming languages
C/C++/OpenMP (RTEMS Source Builder, RSB)
Ada (RSB,
--with-ada)Erlang
Fortran (RSB,
--with-fortran)Python and MicroPython
Parallel languages
Thread synchronization and communication
Locking protocols
Scalable timer and timeout support
Lock-free timestamps (FreeBSD timecounters)
Responsive interrupt management
Link-time configurable schedulers
Fixed-priority
Job-level fixed-priority (EDF)
Constant Bandwidth Server (experimental)
Clustered scheduling (SMP feature)
Focus on link-time application-specific configuration
Linker-set based initialization (similar to global C++ constructors)
Operating system uses fine-grained locking (SMP feature)
Dynamic memory allocators
First-fit (default)
Universal Memory Allocator (UMA , libbsd)
File systems
Device drivers
Termios (serial interfaces)
I2C (Linux user-space API compatible)
SPI (Linux user-space API compatible)
Network stacks (legacy, libbsd, lwIP)
USB stack (libbsd)
SD/MMC card stack (libbsd)
Framebuffer (Linux user-space API compatible, Qt)
Application runs in kernel-space and can access hardware directly
libbsd
Port of FreeBSD user-space and kernel-space components to RTEMS
Easy access to FreeBSD software for RTEMS
Support to stay in synchronization with FreeBSD
1.3. Ecosystem¶
The RTEMS Ecosystem is the collection of tools, packages, code, documentation and online content provided by the RTEMS Project. The ecosystem provides a way to develop, maintain, and use RTEMS. It’s parts interact with the user, the host environment, and each other to make RTEMS accessible, useable and predicable.
The ecosystem is for users, developers and maintainers and it is an ongoing effort that needs your help and support. The RTEMS project is always improving the way it delivers the kernel to you and your feedback is important so please join the mailing lists and contribute back comments, success stories, bugs and patches.
What the RTEMS project describes here to develop, maintain and use RTEMS does not dictate what you need to use in your project. You can and should select the work-flow that best suites the demands of your project and what you are delivering.
1.3.1. Rational¶
RTEMS is complex and the focus of the RTEMS Ecosystem is to simplify the complexity for users by providing a stable documented way to build, configure and run RTEMS. RTEMS is more than a kernel running real-time applications on target hardware, it is part of a project’s and therefore team’s workflow and every project and team is different.
RTEMS’s ecosystem does not mandate a way to work. It is a series of parts, components, and items that are used to create a suitable development environment to work with. The processes explained in this manual are the same things an RTEMS maintainer does to maintain the kernel or an experienced user does to build their production system. It is important to keep this in mind when working through this manual. We encourage users to explore what can be done and to discover ways to make it fit their needs. The ecosystem provided by the RTEMS Project will not install in a single click of a mouse because we want users to learn the parts they will come to depend on as their project’s development matures.
The RTEMS Ecosystem provides a standard interface that is the same on all supported host systems. Standardizing how a user interacts with RTEMS is important and making that experience portable is also important. As a result the ecosystem is documented at the command line level and we leave GUI and IDE integration for users and integrators.
Standardizing the parts and how to use them lets users create processes and procedures that are stable over releases. The RTEMS Ecosystem generates data that can be used to audit the build process so their configuration can be documented.
The ecosystem is based around the source code used in the various parts, components and items of the RTEMS development environment. A user can create an archive of the complete build process including all the source code for long term storage. This is important for projects with a long life cycle.
1.3.2. Open Source¶
RTEMS is an open source operating system and an open source project and this extends to the ecosystem. We encourage users to integrate the processes to build tools, the kernel and any third-party libraries into their project’s configuration management processes.
All the parts that make up the ecosystem are open source. The ecosystem uses a package’s source code to create an executable on a host so when an example RTEMS executable is created and run for the first time the user will have built every tool as well as the executable from source. The RTEMS Project believes the freedom this gives a user is as important as the freedom of having access to the source code for a package.
1.3.3. Deployment¶
The RTEMS Project provides the ecosystem as source code that users can download to create personalised development environments. The RTEMS Project does not provide packaging and deployment for a specific host environment, target architecture or BSP. The RTEMS Project encourages users and organizations to fill this role for the community. The RTEMS Source Builder provides some aid to build and deploy tool binaries.
1.4. Real-time Application Systems¶
Real-time application systems are a special class of computer applications. They have a complex set of characteristics that distinguish them from other software problems. Generally, they must adhere to more rigorous requirements. The correctness of the system depends not only on the results of computations, but also on the time at which the results are produced. The most important and complex characteristic of real-time application systems is that they must receive and respond to a set of external stimuli within rigid and critical time constraints referred to as deadlines. Systems can be buried by an avalanche of interdependent, asynchronous or cyclical event streams.
Deadlines can be further characterized as either hard or soft based upon the value of the results when produced after the deadline has passed. A deadline is hard if the results have no value after the deadline has passed, or a catastrophic event results from their intended use if not completed on time. In contrast, results produced after a soft deadline may still have some value.
Another distinguishing requirement of real-time application systems is the ability to coordinate or manage a large number of concurrent activities. Since software is a synchronous entity, this presents special problems. One instruction follows another in a repeating synchronous cycle. Even though mechanisms have been developed to allow for the processing of external asynchronous events, the software design efforts required to process and manage these events and tasks are growing more complicated.
The design process is complicated further by spreading this activity over a set of processors instead of a single processor. The challenges associated with designing and building real-time application systems become very complex when multiple processors are involved. New requirements such as interprocessor communication channels and global resources that must be shared between competing processors are introduced. The ramifications of multiple processors complicate each and every characteristic of a real-time system.
1.5. Real-time Executive¶
Fortunately, real-time operating systems, or real-time executives, serve as a cornerstone on which to build the application system. A real-time multitasking executive allows an application to be cast into a set of logical, autonomous processes or tasks which become quite manageable. Each task is internally synchronous, but different tasks execute independently, resulting in an asynchronous processing stream. Tasks can be dynamically paused for many reasons resulting in a different task being allowed to execute for a period of time. The executive also provides an interface to other system components such as interrupt handlers and device drivers. System components may request the executive to allocate and coordinate resources, and to wait for and trigger synchronizing conditions. The executive system calls effectively extend the CPU instruction set to support efficient multitasking. By causing tasks to travel through well-defined state transitions, system calls permit an application to demand-switch between tasks in response to real-time events.
By properly grouping stimuli responses into separate tasks a system can now asynchronously switch between independent streams of execution. This allows the system to directly respond to external stimuli as they occur, as well as meet critical performance specifications that are typically measured by guaranteed response time and transaction throughput. The multiprocessor extensions of RTEMS provide the features necessary to manage the extra requirements introduced by a system distributed across several processors. It removes the physical barriers of processor boundaries from the world of the system designer, enabling more critical aspects of the system to receive the required attention. Such a system, based on an efficient real-time, multiprocessor executive, is a more realistic model of the outside world or environment for which it is designed. As a result, the system will always be more logical, efficient, and reliable.
By using the directives provided by RTEMS, the real-time applications developer is freed from the problem of controlling and synchronizing multiple tasks and processors. In addition, one need not develop, test, debug, and document routines to manage memory, pass messages, or provide mutual exclusion. The developer is then able to concentrate solely on the application. By using standard software components, the time and cost required to develop sophisticated real-time applications are significantly reduced.
- 1
The goal is to use the BSD 2-Clause license for new code or code those copyright holder agreed to a license change, see #3053 for the details.
- 2
See #2832.
- 3
Thread-local storage requires some support by the tool chain and the RTEMS architecture support, e.g. context-switch code. It is supported at least on ARM, PowerPC, RISC-V, SPARC and m68k. Check the RTEMS CPU Architecture Supplement if it is supported.
2. Quick Start¶
Follow the sections of this chapter step by step to get started developing applications on top of RTEMS.
2.1. Preparation¶
You need to perform some basic preparation to get started with RTEMS development. You need tools from your host’s operating system to build the RTEMS tool suite from source. The RTEMS tools you build are used to build the Board Support Package (BSP) libraries for your target hardware from source. The BSP libraries contain the RTEMS operating system. This is not a one-click installation process, but there are good reasons to build everything from source.
During this Quick Start guide you will:
Select a suitable place to install RTEMS.
Select if you download all the source code before you start building RTEMS or the source is downloaded on demand as it is needed. If you do not have a reliable internet connection we recommend you download all the source before starting a build.
Build a tool suite.
Build and test a BSP.
Optionally build additional packages.
Alternatively you can build a BSP as a package using the RSB. This is covered in Build an RSB Package
2.1.1. Host Computer¶
The host computer is a computer you use to develop applications. It runs all your tools, editors, documentation viewers, etc. You need a native C, C++, and Python development environment. Please make sure you can build native C/C++ applications on your host computer. You must be able to build native Python C modules as some RTEMS tools contain these modules. Usually, you have to install a Python development package for this. Please have a look at the Host Computer chapter for the gory details. In particular Microsoft Windows users should do this.
2.1.2. Selecting a BSP¶
If you are new to RTEMS and you are looking to try RTEMS then the best suited Board Support Package (BSP) is the SPARC ERC32 (erc32). The SPARC ERC32 BSP has a robust simulator that runs the example and test executables on your host computer. This Quick Start guide will build the erc32 BSP and run RTEMS tests executables in the simulator. The ERC32 BSP is a SPARC architecture BSP so the tool suite name is sparc-rtems5.
If you are looking for a hardware target to run RTEMS on we recommend the BeagleBone Black (beagleboneblack) BSP. The BeagleBone Black support includes the RTEMS BSD Library (libbsd) and networking. The BeagleBone Black BSP is an ARM architecture BSP so the tool suite name is arm-rtems5.
2.1.3. Selecting a Version of RTEMS¶
In the examples of this manual we will often refer to a specific version of RTEMS, which will usually be the version that accompanied the publication of this documentation manual. That may not be the appropriate version for you to use, for example, it may be too old (or too new) depending on what you are trying to do. If you’re not sure what version to use, we generally recommend using the most recent release or the development head (master), and you may want to consult with the same version of the documentation. We hope that newer is better.
An RTEMS release involves the creation of a single downloadable file, normally a compressed tarball, that packages the source of all the repositories in a state consistent with the time the release is created. A release branch is a git branch pushed to the repositories named with the numeric identifier of the branch. A release branch release is a git tag on a release branch with the tags pushed to the repositories.
Numbering for RTEMS versions beginning with RTEMS 5 uses a format as follows. The master branch has the version N.0.0 with N being the next major release number. The first release of this series has the version number N.1.0. and there is exactly one commit with this version number in the corresponding repository. The first bugfix release (minor release) of this series will have the version number N.2.0. The release branch will have the version number N.M.1 with M being the last minor release of this series.
For example: + 5.0.0 is the version number of the development master for the 5 series. + 5.1.0 is the first release of the 5 series. + 5.1.1 is the version number of the 5 series release branch right after
the 5.1.0 release until 5.2.0 is released.
5.2.0 is the first bugfix release of the 5 series
5.2.1 is the version number of the 5 series release branch right after the 5.2.0 release until 5.3.0 is released.
6.0.0 is the version number of the development master for the 6 series.
RTEMS development tools use N as the version number and are expected to work with all releases and the release branch of the N series. So to build tools for compiling RTEMS version number 5.1.0 for SPARC use sparc-rtems5. Despite the number not increasing, the tools may change within a release branch, for example the tools packaged with 5.1.1 still use the sparc-rtems5 moniker, but are likely not the same as the tools used in version 5.1.0. This tool mismatch can be a source of confusion. Be sure to use the toolchain that matches your release.
2.2. Choose an Installation Prefix¶
You will see the term prefix referred to throughout this documentation and in a wide number of software packages you can download from the internet. It is also used in the GNU Coding Standard. A prefix is the path on your host computer a software package is installed under. Packages that have a prefix will place all parts under the prefix path. Packages for your host computer typically use a default prefix of /usr/local on FreeBSD and Linux.
You have to select a prefix for your installation. You will build and install the RTEMS tool suite, an RTEMS kernel for a BSP, and you may build and install third party libraries. You can build all the parts as a stack with a single prefix or you can separate various parts by providing different prefixes to each part as it is built. Using separate prefixes is for experienced RTEMS users.
Do not select a prefix that is under the top of any of the source trees. The prefix collects the install output of the various build steps you take in this guide and need to be kept separate from the sources used.
The RTEMS tool suite consists of a cross tool chain (Binutils, GCC, GDB, Newlib, etc.) for your target architecture and RTEMS tools provided by the RTEMS Project. The RTEMS Tools are a toolkit that help create the RTEMS ecosystem and help support the building of embedded real-time applications and systems.
You build and install the tool suite with the RTEMS Source Builder (RSB). By default, the RSB will start the prefix path with a host operating system specific path plus rtems, and the RTEMS version, e.g. /opt/rtems/5 on Linux, and /usr/local/rtems/5 on FreeBSD and macOS. Placing the RTEMS version number in the path lets you manage and migrate RTEMS versions as they are released.
It is strongly recommended to run the RSB as a normal user and not with root privileges (also known as super user or Administrator). You have to make sure that your normal user has sufficient privileges to create files and directories under the prefix. For example, you can create a directory /opt/rtems and give it to a developer group with read, write, and execute permissions. Alternatively, you can choose a prefix in your home directory, e.g. $HOME/rtems/5 or with a project-specific component $HOME/project-x/rtems/5. For more ideas, see the project sandboxing section. In this quick start chapter, we will choose $HOME/quick-start/rtems/5 for the RTEMS tool suite prefix.
Warning
The prefix must not contain space characters.
2.3. Obtain the Sources¶
You have considered and chosen a suitable installation prefix in the previous section. We have chosen $HOME/quick-start/rtems/5 as the installation prefix. We will show how to use a released version of RTEMS and then as an alternative we will show you using the RSB Git repository. Consider using a Git clone if you wish to make contributions to the RTEMS Project.
You need the RTEMS Source Builder (RSB) to work with RTEMS and we prefer you use a released version. A released version of the RSB downloads all source code from the RTEMS servers. Each release archives all the referenced source providing long term stability as changes in upstream projects do not effect a release’s build.
You will need approximately 1.5G bytes of disk space to build the tools, RTEMS kernel, network stack and 3rd party packages for the ERC32 BSP.
2.3.1. Releases¶
You can download the source archives for a released RTEMS version from RTEMS’ servers. Releases can be view at https://ftp.rtems.org/pub/rtems/releases with the releases listed as a series under a release’s major number. For RTEMS 5.1.0 the release series is 5 and the release path is https://ftp.rtems.org/pub/rtems/releases/5/5.1.0.
To work with the archives of a released RTEMS version, simply replace the version number 5 used throughout this chapter with the version number you selected, e.g. sparc-rtems4.11, sparc-rtems6, and so on.
Download and unpack using the curl and tar command with these commands:
mkdir -p $HOME/quick-start/src
cd $HOME/quick-start/src
curl https://ftp.rtems.org/pub/rtems/releases/5/5.1.0/rtems-source-builder-5.1.0.tar.xz | tar xJf -
If curl does not work consider using wget or a browser.
The RSB is unpacked under the path rtems-source-builder-5.1.0. Rename this to rsb to get shorter paths during the tool suite build. To do this run these commands:
cd $HOME/quick-start/src
mv rtems-source-builder-5.1.0 rsb
If you wish to build the RTEMS kernel from source obtain the RTEMS kernel sources:
cd $HOME/quick-start/src
curl https://ftp.rtems.org/pub/rtems/releases/5/5.1.0/rtems-5.1.0.tar.xz | tar xJf -
2.3.2. Git¶
Alternatively, clone the Git repositories into $HOME/quick-start/src.
A Git repository clone gives you some flexibility with the added complexity of needing to use a Git branch to build a released version. With Git you can switch between branches to try out different RTEMS versions and you have access to the RTEMS source history. The RTEMS Project welcomes contributions. The Git repositories enable you to easily create patches and track local changes.
You can clone the Git repository to get all versions of RTEMS including the development head. Release branches in Git are kept stable however they may differ from a release’s source archive.
mkdir -p $HOME/quick-start/src
cd $HOME/quick-start/src
git clone git://git.rtems.org/rtems-source-builder.git rsb
git clone git://git.rtems.org/rtems.git
The rsb repository clone contains the RTEMS Source Builder (RSB). We clone it into rsb to get shorter paths during the tool suite build. The rtems repository clone contains the RTEMS sources. These two repositories are enough to get started. There are more repositories available.
2.3.3. Offline Download¶
If you have limited Internet access you can download the source before you start building. If you are permanently connected to the Internet you do not need to do this and the sources will be automatically download on demand when needed.
Once the sources have been downloaded you could disconnect your host computer from the Internet. It is no longer required to work with RTEMS. To download the sources to build the ERC 32 BSP before building run the following commands:
cd $HOME/quick-start/src/rsb/rtems
../source-builder/sb-set-builder --source-only-download 5/rtems-sparc
This command should output something like this (omitted lines are denoted by ...):
RTEMS Source Builder - Set Builder, 5.1.0
Build Set: 5/rtems-sparc
...
download: https://ftp.rtems.org/pub/rtems/releases/5/5.1.0/5.1.0/sources/gcc-7.5.0.tar.xz -> sources/gcc-7.5.0.tar.xz
...
Build Sizes: usage: 0.000B total: 143.814MB (sources: 143.793MB, patches: 21.348KB, installed 0.000B)
Build Set: Time 0:05:52.617958
If you encounter errors, check your internet connection, firewall settings, virus scanners and the availability of the download servers.
2.4. Install the Tool Suite¶
You have chosen an installation prefix, the BSP to build, the tool’s architecure and prepared the source for the RSB in the previous sections. We have chosen $HOME/quick-start/rtems/5 as the installation prefix, the erc32 BSP and the SPARC architecture name of sparc-rtems5, and unpacked the RSB source in $HOME/quick-start/src.
The tool suite for RTEMS and the RTEMS sources are tightly coupled. For example, do not use a RTEMS version 5 tool suite with RTEMS version 4.11 sources and vice versa.
Build and install the tool suite:
cd $HOME/quick-start/src/rsb/rtems
../source-builder/sb-set-builder --prefix=$HOME/quick-start/rtems/5 5/rtems-sparc
This command should output something like this (omitted lines are denoted by …). The build host appears as part of the name of the package being built. The name you see may vary depending on the host you are using:
RTEMS Source Builder - Set Builder, 5.1.0
Build Set: 5/rtems-sparc
...
config: tools/rtems-binutils-2.34.cfg
package: sparc-rtems5-binutils-2.34-x86_64-freebsd12.1-1
building: sparc-rtems5-binutils-2.34-x86_64-freebsd12.1-1
sizes: sparc-rtems5-binutils-2.34-x86_64-freebsd12.1-1: 305.866MB (installed: 29.966MB)
cleaning: sparc-rtems5-binutils-2.34-x86_64-freebsd12.1-1
reporting: tools/rtems-binutils-2.34.cfg -> sparc-rtems5-binutils-2.34-x86_64-freebsd12.1-1.txt
reporting: tools/rtems-binutils-2.34.cfg -> sparc-rtems5-binutils-2.34-x86_64-freebsd12.1-1.xml
config: tools/rtems-gcc-7.5.0-newlib-fbaa096.cfg
package: sparc-rtems5-gcc-7.5.0-newlib-fbaa096-x86_64-freebsd12.1-1
building: sparc-rtems5-gcc-7.5.0-newlib-fbaa096-x86_64-freebsd12.1-1
....
Build Sizes: usage: 5.684GB total: 1.112GB (sources: 143.803MB, patches: 21.348KB, installed 995.188MB)
Build Set: Time 0:21:35.626294
Once the build has successfully completed you can check if the cross C compiler works with the following command:
$HOME/quick-start/rtems/5/bin/sparc-rtems5-gcc --version
This command should output something like below. The version informtion helps you to identify the exact sources used to build the cross compiler of your RTEMS tool suite. In the output you see the version of RTEMS or the hash from the RSB repository if you are building using a Git repository clone. The Newlib hash is the version of Newlib in the RTEMS’s github sourceware-mirror-newlib-cygwin repository. The sources and patches directories created by the RSB contain all the source code used.
sparc-rtems5-gcc (GCC) 7.5.0 20191114 (RTEMS 5, RSB 5.1.0, Newlib fbaa096)
Copyright (C) 2017 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
Add --verbose to the GCC command for the the verbose version details.
2.4.1. Need for RTEMS-Specific Cross-Compiler¶
New users are often confused as to why they cannot use their distribution’s cross-compiler for their target on RTEMS, e.g., the riscv64-linux-gnu or the arm-none-eabi-gcc. Below mentioned are some of the reasons for using the RTEMS cross-compiler.
- Correct configuration of Newlib
Newlib is a C standard library implementation intended for use on embedded systems. Most of the POSIX and libc support for RTEMS is derived from Newlib. The RTEMS cross-compiler configures Newlib correctly for RTEMS.
- Threading in GCC support libraries
Several threading packages in GCC such as Go threads (libgo), OpenMP (libgomp), and OpenACC need to be customized according to RTEMS. This is done by the RTEMS specific cross-compiler.
- Provide preprocessor define __rtems__
The
__rtems__preprocessor define is used to provide conditional code compilation in source files that are shared with other projects e.g. in Newlib or imported code from FreeBSD.- Multilib variants to match the BSP
RTEMS configures GCC to create separate runtime libraries for each supported instruction set, floating point unit, vector unit, word size (e.g. 32-bit and 64-bit), endianness, ABI, processor errata workarounds, and so on in the architecture. These libraries are termed as Multilib variants. Multilib variants to match the BSP are set by selecting a specific set of machine options using the RTEMS cross-compiler.
2.5. Bootstrap the RTEMS Sources¶
You installed the tool suite in your installation prefix and made ready the source for two RTEMS source packages in the previous sections. We installed the tool suite in $HOME/quick-start/rtems/5 and unpacked the RSB source in $HOME/quick-start/src.
You only need to bootstrap the RTEMS sources if you have used Git to get the sources. If you use source archives of a released RTEMS version you can skip this section and move to Build a Board Support Package (BSP).
Before you can build a Board Support Package (BSP) for your target hardware from Git cloned RTEMS sources, you have to bootstrap the build system in the RTEMS sources. This is only necessary if you use a Git repository clone of the RTEMS sources. You have to do this after a fresh repository clone and sometimes after build system file updates (e.g. after a git pull). If you are not a build system expert, then do the bootstrap after each update of build system files. This is a bit annoying, but improving the build system is a complex and time consuming undertaking. Feel free to help the RTEMS Project to improve it. For the bootstrap it is important that the right version of Autotools (autoconf and automake) are in your $PATH. The right version of Autotools is shipped with the RTEMS tool suite you already installed. Set the path to the tool suite installed under your selected prefix:
export PATH=$HOME/quick-start/rtems/5/bin:"$PATH"
Change into the RTEMS source tree to bootstrap the build system:
cd $HOME/quick-start/src/rtems
./rtems-bootstrap
This command should output something like this (omitted lines are denoted by ...):
RTEMS Bootstrap, 1.0
1/122: autoreconf: configure.ac
2/122: autoreconf: testsuites/configure.ac
3/122: autoreconf: testsuites/fstests/configure.ac
4/122: autoreconf: testsuites/smptests/configure.ac
5/122: autoreconf: testsuites/psxtests/configure.ac
6/122: autoreconf: testsuites/mptests/configure.ac
...
121/122: autoreconf: c/src/lib/libbsp/lm32/milkymist/configure.ac
122/122: autoreconf: c/src/make/configure.ac
Bootstrap time: 0:00:46.404643
2.6. Build a Board Support Package (BSP)¶
You installed the tool suite in your installation prefix, made ready the source for two RTEMS source packages and if you are using a Git clone bootstrapped the RTEMS sources in the previous sections. We installed the tool suite in $HOME/quick-start/rtems/5 and unpacked the source in $HOME/quick-start/src.
You are now able to build Board Support Packages (BSPs) for all architectures you have an installed RTEMS tool suite. To build applications on top of RTEMS, you have to build and install a BSP for your target hardware. To select a proper BSP for your target hardware consult the BSPs chapter. We select the erc32 BSP. The erc32 BSP uses approximately 2.3G bytes of disk space when the testsuite is built and 44M bytes of space when installed.
We will first show how to build the BSP using the RSB and then we will show how to build the same BSP manually. You only need to use one of the listed methods to build the BSP.
In the output in this section the base directory $HOME/quick-start was replaced by $BASE.
2.6.1. RSB BSP Build¶
The RSB build of RTEMS does not use the RTEMS source we made ready. It uses the RSB source you downloaded in a previous section. If you are using a release RSB source archive the BSP built is the released kernel image. If you are using a Git clone of the RSB the BSP will be version referenced in the RSB clone.
To build the BSP with all the tests run this command:
cd $HOME/quick-start/src/rsb/rtems
../source-builder/sb-set-builder --prefix=$HOME/quick-start/rtems/5 \
--target=sparc-rtems5 --with-rtems-bsp=erc32 --with-rtems-tests=yes 5/rtems-kernel
This command should output something like this:
RTEMS Source Builder - Set Builder, 5.1.0
Build Set: 5/rtems-kernel
config: tools/rtems-kernel-5.cfg
package: sparc-rtems5-kernel-erc32-1
building: sparc-rtems5-kernel-erc32-1
sizes: sparc-rtems5-kernel-erc32-1: 2.279GB (installed: 44.612MB)
cleaning: sparc-rtems5-kernel-erc32-1
reporting: tools/rtems-kernel-5.cfg -> sparc-rtems5-kernel-erc32-1.txt
reporting: tools/rtems-kernel-5.cfg -> sparc-rtems5-kernel-erc32-1.xml
installing: sparc-rtems5-kernel-erc32-1 -> $BASE/
cleaning: sparc-rtems5-kernel-erc32-1
Build Set: Time 0:03:09.896961
The RSB BSP build can be customised with following RSB command line options:
--with-rtems-tests:Build the test suite. If
yesis provided all tests in the testsuite are build. Ifnois provided no tests are built and ifsamplesis provided only the sample executables are built, e.g.--with-rtems-tests=yes. The test executables are install under the BSP in thetestsdirectory and you can execute them with the tester and run command.--with-rtems-smp:Build with SMP support. The BSP has to have SMP support or this option will fail with an error.
--with-rtems-legacy-network:Build the legacy network software. We recommend you use the current network support in the RTEMS BSP Library (libbsd) unless you need to maintain a legacy product. Do not use the legacy networking software for new developments.
--with-rtems-bspopts:Build the BSP with BSP specific options. This is an advanced option. Please refer to the BSP specific details in the Board Support Packages (BSPs) of this manual or the BSP source code in the RTEMS source directory. To supply a list of options quote then list with
", e.g.--with-rtems-bspopts="BSP_POWER_DOWN_AT_FATAL_HALT=1"
If you have built a BSP with the RSB, you can move on to Test a Board Support Package (BSP).
2.6.2. Manual BSP Build¶
We manually build the BSP in four steps. The first step is to create a build directory. It must be separate from the RTEMS source directory. We use $HOME/quick-start/build/b-erc32.
mkdir -p $HOME/quick-start/build/b-erc32
The second step is to set your path. Prepend the RTEMS tool suite binary directory to your $PATH throughout the remaining steps. Run the command:
export PATH=$HOME/quick-start/rtems/5/bin:"$PATH"
Check your installed tools can be found by running:
command -v sparc-rtems5-gcc && echo "found" || echo "not found"
The output should be:
found
If not found is printed the tools are not correctly installed or the path has not been correctly set. Check the contents of the path $HOME/quick-start/rtems/5/bin manually and if sparc-rtems5-gcc is present the path is wrong. If the file cannot be found return to Install the Tool Suite and install the tools again.
The third step is to configure the BSP. There are various configuration options available. Some configuration options are BSP-specific.
cd $HOME/quick-start/build/b-erc32
$HOME/quick-start/src/rtems/configure \
--prefix=$HOME/quick-start/rtems/5 \
--enable-maintainer-mode \
--target=sparc-rtems5 \
--enable-rtemsbsp=erc32 \
--enable-tests
This command should output something like this (omitted lines are denoted by ...):
checking for gmake... gmake
checking for RTEMS Version... 5.0.0
checking build system type... x86_64-unknown-freebsd12.0
checking host system type... x86_64-unknown-freebsd12.0
checking target system type... sparc-unknown-rtems5
...
config.status: creating Makefile
target architecture: sparc.
available BSPs: erc32.
'gmake all' will build the following BSPs: erc32.
other BSPs can be built with 'gmake RTEMS_BSP="bsp1 bsp2 ..."'
config.status: creating Makefile
Building the BSP is the forth step.
cd $HOME/quick-start/build/b-erc32
make
This command should output something like this (omitted lines are denoted by …).
Configuring RTEMS_BSP=erc32
checking for gmake... gmake
checking build system type... x86_64-unknown-freebsd12.0
checking host system type... sparc-unknown-rtems5
checking rtems target cpu... sparc
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for sparc-rtems5-strip... sparc-rtems5-strip
checking for a thread-safe mkdir -p... $BASE/src/rtems/c/src/../../install-sh -c -d
checking for gawk... no
checking for mawk... no
checking for nawk... nawk
checking whether gmake sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... yes
checking for RTEMS_BSP... erc32
checking whether CPU supports libposix... yes
configure: setting up make/custom
configure: creating make/erc32.cache
gmake[3]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32'
...
sparc-rtems5-gcc -mcpu=cypress -O2 -g -ffunction-sections -fdata-sections -Wall -Wmissing-prototypes -Wimplicit-function-declaration -Wstrict-prototypes -Wnested-externs -B./../../lib/libbsp/sparc/erc32 -B$BASE/src/rtems/bsps/sparc/erc32/start -specs bsp_specs -qrtems -L./../../cpukit -L$BASE/src/rtems/bsps/sparc/shared/start -Wl,--wrap=printf -Wl,--wrap=puts -Wl,--wrap=putchar -Wl,--gc-sections -o spwkspace.exe spwkspace/spwkspace-init.o ./../../lib/libbsp/sparc/erc32/librtemsbsp.a ./../../cpukit/librtemscpu.a
gmake[5]: Leaving directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32/testsuites/sptests'
gmake[4]: Leaving directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32/testsuites'
gmake[3]: Leaving directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32'
gmake[2]: Leaving directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32'
gmake[1]: Leaving directory '$BASE/build/b-erc32/sparc-rtems5/c'
gmake[1]: Entering directory '$BASE/build/b-erc32'
gmake[1]: Nothing to be done for 'all-am'.
gmake[1]: Leaving directory '$BASE/build/b-erc32'
The last step is to install the BSP.
cd $HOME/quick-start/build/b-erc32
make install
This command should output something like this (omitted lines are denoted by …). In this output the base directory $HOME/quick-start was replaced by $BASE.
Making install in sparc-rtems5/c
gmake[1]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c'
Making install in .
gmake[2]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c'
gmake[3]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c'
gmake[3]: Nothing to be done for 'install-exec-am'.
gmake[3]: Nothing to be done for 'install-data-am'.
gmake[3]: Leaving directory '$BASE/build/b-erc32/sparc-rtems5/c'
gmake[2]: Leaving directory '$BASE/build/b-erc32/sparc-rtems5/c'
Making install in erc32
gmake[2]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32'
gmake[3]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32'
Making install-am in .
Making install-am in cpukit
gmake[4]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32/cpukit'
gmake[5]: Entering directory '$BASE/build/b-erc32/sparc-rtems5/c/erc32/cpukit'
gmake[5]: Nothing to be done for 'install-exec-am'.
$BASE/src/rtems/c/src/../../cpukit/../install-sh -c -d '$BASE/rtems/5/sparc-rtems5/erc32/lib/include'
...
$BASE/src/rtems/make/Templates/Makefile.lib '$BASE/rtems/5/share/rtems5/make/Templates'
$BASE/src/rtems/./install-sh -c -d '$BASE/rtems/5/make/custom'
/usr/bin/install -c -m 644 $BASE/src/rtems/make/custom/default.cfg '$BASE/rtems/5/make/custom'
gmake[2]: Leaving directory '$BASE/build/b-erc32'
gmake[1]: Leaving directory '$BASE/build/b-erc32'
2.7. Test a Board Support Package (BSP)¶
You built a BSP with tests in the previous section. We built the erc32 BSP in $HOME/quick-start/build/b-erc32.
You should run the RTEMS test suite on your target hardware. The RTEMS Project provides some support to do this, see the Testing chapter for the details.
On the erc32 BSP we selected for this quick start chapter this is easy. Just run this command:
cd $HOME/quick-start/build/b-erc32
rtems-test --rtems-bsp=erc32-sis --rtems-tools=$HOME/quick-start/rtems/5 .
This command should output something like this (omitted lines are denoted by …). In this output the base directory $HOME/quick-start was replaced by $BASE.
RTEMS Testing - Tester, 5.1.0
Command Line: $BASE/rtems/5/bin/rtems-test --rtems-bsp=erc32-sis --rtems-tools=$BASE/rtems/5 .
Python: 2.7.15 (default, Jan 10 2019, 01:14:47) [GCC 4.2.1 Compatible FreeBSD Clang 6.0.1 (tags/RELEASE_601/final 335540)]
Host: FreeBSD-12.0-RELEASE-p2-amd64-64bit-ELF (FreeBSD Build_FreeBSD12 12.0-RELEASE-p2 FreeBSD 12.0-RELEASE-p2 GENERIC amd64 amd64)
[ 1/589] p:0 f:0 u:0 e:0 I:0 B:0 t:0 i:0 W:0 | sparc/erc32: dhrystone.exe
...
[589/589] p:574 f:0 u:5 e:0 I:0 B:3 t:0 i:0 W:0 | sparc/erc32: tmtimer01.exe
Passed: 580
Failed: 0
User Input: 5
Expected Fail: 0
Indeterminate: 0
Benchmark: 3
Timeout: 1
Invalid: 0
Wrong Version: 0
Wrong Build: 0
Wrong Tools: 0
------------------
Total: 589
User Input:
monitor.exe
termios.exe
top.exe
fileio.exe
capture.exe
Benchmark:
whetstone.exe
linpack.exe
dhrystone.exe
Timeouts:
pppd.exe
Average test time: 0:00:00.437773
Testing time : 0:04:17.848557
2.8. Build Your Application¶
You tested a BSP in the previous section. We built the erc32 BSP and it is installed under $HOME/quick-start/rtems/5.
We will now create a simple Hello World application with a Git repository and using the Waf build system.
The application is be created in $HOME/quick-start/app/hello.
In the output in this section the base directory $HOME/quick-start was replaced by $BASE.
The steps in this section assume you are in the directory $HOME/quick-start/app/hello after the first step changes to it.
Setup the application work space. Create a new Git repository, download the Waf build system, and the RTEMS Waf.
Create the application directory and change into it:
mkdir -p $HOME/quick-start/app/hello
cd $HOME/quick-start/app/hello
Download the Waf build system and set it to executable:
curl https://waf.io/waf-2.0.19 > waf
chmod +x waf
Initialise a new Git repository:
git init
Add RTEMS Waf support as a Git sub-module and initialise it:
git submodule add git://git.rtems.org/rtems_waf.git rtems_waf
Create the application source files. Three files are created with an editor of your choice.
First create a C file that configures RTEMS. Using an editor create a file called init.c and copy the following configuration settings:
/*
* Simple RTEMS configuration
*/
#define CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER
#define CONFIGURE_APPLICATION_NEEDS_CONSOLE_DRIVER
#define CONFIGURE_UNLIMITED_OBJECTS
#define CONFIGURE_UNIFIED_WORK_AREAS
#define CONFIGURE_RTEMS_INIT_TASKS_TABLE
#define CONFIGURE_INIT
#include <rtems/confdefs.h>
Create the Hello World application source file. Using an editor create hello.c and copy the follow code:
/*
* Hello world example
*/
#include <rtems.h>
#include <stdlib.h>
#include <stdio.h>
rtems_task Init(
rtems_task_argument ignored
)
{
printf( "\nHello World\n" );
exit( 0 );
}
Finally create the Waf script. Using an editor create wscript and copy the Waf script:
#
# Hello world Waf script
#
from __future__ import print_function
rtems_version = "5"
try:
import rtems_waf.rtems as rtems
except:
print('error: no rtems_waf git submodule')
import sys
sys.exit(1)
def init(ctx):
rtems.init(ctx, version = rtems_version, long_commands = True)
def bsp_configure(conf, arch_bsp):
# Add BSP specific configuration checks
pass
def options(opt):
rtems.options(opt)
def configure(conf):
rtems.configure(conf, bsp_configure = bsp_configure)
def build(bld):
rtems.build(bld)
bld(features = 'c cprogram',
target = 'hello.exe',
cflags = '-g -O2',
source = ['hello.c',
'init.c'])
Configure the application using Waf’s configure command:
./waf configure --rtems=$HOME/quick-start/rtems/5 --rtems-bsp=sparc/erc32
The output will be something close to:
Setting top to : $BASE/app/hello
Setting out to : $BASE/app/hello/build
RTEMS Version : 5
Architectures : sparc-rtems5
Board Support Package (BSP) : sparc-rtems5-erc32
Show commands : no
Long commands : no
Checking for program 'sparc-rtems5-gcc' : $BASE/rtems/5/bin/sparc-rtems5-gcc
Checking for program 'sparc-rtems5-g++' : $BASE/rtems/5/bin/sparc-rtems5-g++
Checking for program 'sparc-rtems5-gcc' : $BASE/rtems/5/bin/sparc-rtems5-gcc
Checking for program 'sparc-rtems5-ld' : $BASE/rtems/5/bin/sparc-rtems5-ld
Checking for program 'sparc-rtems5-ar' : $BASE/rtems/5/bin/sparc-rtems5-ar
Checking for program 'sparc-rtems5-nm' : $BASE/rtems/5/bin/sparc-rtems5-nm
Checking for program 'sparc-rtems5-objdump' : $BASE/rtems/5/bin/sparc-rtems5-objdump
Checking for program 'sparc-rtems5-objcopy' : $BASE/rtems/5/bin/sparc-rtems5-objcopy
Checking for program 'sparc-rtems5-readelf' : $BASE/rtems/5/bin/sparc-rtems5-readelf
Checking for program 'sparc-rtems5-strip' : $BASE/rtems/5/bin/sparc-rtems5-strip
Checking for program 'sparc-rtems5-ranlib' : $BASE/rtems/5/bin/sparc-rtems5-ranlib
Checking for program 'rtems-ld' : $BASE/rtems/5/bin/rtems-ld
Checking for program 'rtems-tld' : $BASE/rtems/5/bin/rtems-tld
Checking for program 'rtems-syms' : $BASE/rtems/5/bin/rtems-syms
Checking for program 'rtems-bin2c' : $BASE/rtems/5/bin/rtems-bin2c
Checking for program 'tar' : /usr/bin/tar
Checking for program 'gcc, cc' : $BASE/rtems/5/bin/sparc-rtems5-gcc
Checking for program 'ar' : $BASE/rtems/5/bin/sparc-rtems5-ar
Checking for program 'g++, c++' : $BASE/rtems/5/bin/sparc-rtems5-g++
Checking for program 'ar' : $BASE/rtems/5/bin/sparc-rtems5-ar
Checking for program 'gas, gcc' : $BASE/rtems/5/bin/sparc-rtems5-gcc
Checking for program 'ar' : $BASE/rtems/5/bin/sparc-rtems5-ar
Checking for c flags '-MMD' : yes
Checking for cxx flags '-MMD' : yes
Compiler version (sparc-rtems5-gcc) : 7.5.0 20191114 (RTEMS 5, RSB 5.1.0, Newlib fbaa096)
Checking for a valid RTEMS BSP installation : yes
Checking for RTEMS_DEBUG : no
Checking for RTEMS_MULTIPROCESSING : no
Checking for RTEMS_NEWLIB : yes
Checking for RTEMS_POSIX_API : yes
Checking for RTEMS_SMP : no
Checking for RTEMS_NETWORKING : no
'configure' finished successfully (0.686s)
Build the application:
./waf
The output will be something close to:
Waf: Entering directory `$BASE/app/hello/build/sparc-rtems5-erc32'
[1/3] Compiling init.c
[2/3] Compiling hello.c
[3/3] Linking build/sparc-rtems5-erc32/hello.exe
Waf: Leaving directory `$BASE/app/hello/build/sparc-rtems5-erc32'
'build-sparc-rtems5-erc32' finished successfully (0.183s)
Run the executable:
$HOME/quick-start/rtems/5/bin/rtems-run --rtems-bsps=erc32-sis build/sparc-rtems5-erc32/hello.exe
The output will be something close to:
RTEMS Testing - Run, 5.1.0
Command Line: $BASE/rtems/5/bin/rtems-run --rtems-bsps=erc32-sis build/sparc-rtems5-erc32/hello.exe
Host: FreeBSD hihi 12.1-RELEASE-p2 FreeBSD 12.1-RELEASE-p2 GENERIC amd64
Python: 3.7.6 (default, Jan 30 2020, 01:18:54) [Clang 6.0.1 (tags/RELEASE_601/final 335540)]
Host: FreeBSD-12.1-RELEASE-p2-amd64-64bit-ELF (FreeBSD hihi 12.1-RELEASE-p2 FreeBSD 12.1-RELEASE-p2 GENERIC amd64 amd64)
SIS - SPARC/RISCV instruction simulator 2.21, copyright Jiri Gaisler 2019
Bug-reports to jiri@gaisler.se
ERC32 emulation enabled
Loaded build/sparc-rtems5-erc32/hello.exe, entry 0x02000000
Hello World
*** FATAL ***
fatal source: 5 (RTEMS_FATAL_SOURCE_EXIT)
fatal code: 0 (0x00000000)
RTEMS version: 5.1.0
RTEMS tools: 7.5.0 20191114 (RTEMS 5, RSB 5.1.0, Newlib fbaa096)
executing thread ID: 0x08a010001
executing thread name: UI1
cpu 0 in error mode (tt = 0x101)
107883 0200b6c0: 91d02000 ta 0x0
Run time : 0:00:01.011474
Commit the application to the repository:
git add init.c hello.c wscript
git commit -m "My first RTEMS application."
2.9. Build an RSB Package¶
This section describes how to build an RTEMS package using the RSB. Before we start to build a package with the RSB you need to complete these steps:
Return to here once you have completed these steps.
You have chosen an installation prefix, the BSP to build, the tool’s architecure and prepared the source for the RSB in the previous sections. We have chosen $HOME/quick-start/rtems/5 as the installation prefix, the erc32 BSP and the SPARC architecture name of sparc-rtems5, and unpacked the RSB source in $HOME/quick-start/src.
You are now able to build BSP Packages or 3rd party libraries of code if you have built a BSP.
2.9.1. RTEMS Packages¶
RTEMS Packages are source packages the RSB build to run on RTEMS. An installed package is a set of header files and libraries. Your application include the packages header files to make calls to the package’s code and include the libraries in it’s linker options.
RTEMS packages can be part of the RTEMS Project or they can be external packages from 3rd parties. RTEMS Project packages include the BSPs and BSD Library package called libbsd. External 3rd party packages include networking such has curl or libcurl to graphics libraries.
Packages can depend on other packages and need to be build in the corret order. For example the FreeBSD Library package depends on the BSP package and a 3rd party library such as curl depends on the FreeBSD Library package. We call this layering a vertical software stack.
RTEMS applications are cross-compiled and this adds complexity when building libraries of code. RTEMS Packages build with the RSB manage this complexity for you.
Package are libraries so they will not be linked into your application until you make calls to the the code and add the library to your application’s linker command line.
2.9.2. BSP Stack Build¶
A BSP stack build is a single command that uses the RSB to build a BSP software stack of:
Tool suite
BSP
Packages
The packages built depend on the BSP and the default will build all packages for a BSP.
cd $HOME/quick-start/src/rsb/rtems
../source-builder/sb-set-builder --prefix=$HOME/quick-start/rtems/5 \
--with-rtems-tests=yes 5/bsps/erc32
This command should output something like this:
RTEMS Source Builder - Set Builder, 5.1.0
Build Set: 5/bsps/erc32
Build Set: 5/rtems-sparc.bset
Build Set: 5/rtems-autotools.bset
Build Set: 5/rtems-autotools-internal.bset
config: tools/rtems-autoconf-2.69-1.cfg
package: autoconf-2.69-x86_64-freebsd12.1-1
building: autoconf-2.69-x86_64-freebsd12.1-1
sizes: autoconf-2.69-x86_64-freebsd12.1-1: 7.505MB (installed: 0.000B)
...
building: protobuf-2.6.1-sparc-rtems5-1
sizes: protobuf-2.6.1-sparc-rtems5-1: 228.079MB (installed: 84.408MB)
cleaning: protobuf-2.6.1-sparc-rtems5-1
reporting: net/protobuf-2.6.1-1.cfg -> protobuf-2.6.1-sparc-rtems5-1.txt
reporting: net/protobuf-2.6.1-1.cfg -> protobuf-2.6.1-sparc-rtems5-1.xml
staging: protobuf-2.6.1-sparc-rtems5-1 -> $HOME/quick-start/src/rsb/rtems/build/tmp/sb-500-staging
cleaning: protobuf-2.6.1-sparc-rtems5-1
Build Set: Time 0:00:23.564992
Build Set: Time 0:02:27.380299
installing: 5/bsps/erc32 -> $HOME/quick-start/rtems/
clean staging: 5/bsps/erc32
Staging Size: 1.372GB
Build Set: Time 0:24:17.83979
The RSB BSP build can be customised with following RSB command line options:
--with-rtems-tests:Build the test suite. If
yesis provided all tests in the testsuite are build. Ifnois provided no tests are built and ifsamplesis provided only the sample executables are built, e.g.--with-rtems-tests=yes.--with-rtems-smp:Build with SMP support. The BSP has to have SMP support or this option will fail with an error.
--with-rtems-bspopts:Build the BSP with BSP specific options. This is an advanced option. Please refer to the BSP specific details in the Board Support Packages (BSPs) of this manual or the BSP source code in the RTEMS source directory. To supply a list of options quote then list with
", e.g.--with-rtems-bspopts="BSP_POWER_DOWN_AT_FATAL_HALT=1"
Only a limited number of BSPs have RSB support to build as a software stack. To see which BSPs are supported run this command:
cd $HOME/quick-start/src/rsb/rtems
../source-builder/sb-set-builder --list-bsets | grep bsps
2.9.3. Package Build¶
Packages are built using RSB build sets. A build set is a set of builds need to build a packages. The build steps can be dependencies a package has or it could be a stack of software to provide specific functionality, i.e. a build set can be a list of build sets. To view the avaliable build sets run this command:
cd $HOME/quick-start/src/rsb/rtems
../source-builder/sb-set-builder --list-bsets
RTEMS package naming is based on the naming FreeBSD uses in its ports collection.
This Quick Start Guide will build the BSD Library or 5/rtems-libbsd.
An RTEMS package is hosted on RTEMS so the tool suite name needs to be supplied using the --host option, e.g. --host=sparc-rtem5. The BSP needs to be provided using the --with-rtems-bsp option, e.g. --with-rtems-bsp=erc32. The commands to build libbsd for the erc32 BSP are:
cd $HOME/quick-start/src/rsb/rtems
../source-builder/sb-set-builder --prefix=$HOME/quick-start/rtems/5 \
--host=sparc-rtems5 --with-rtems-bsp=erc32 5/rtems-libbsd
This command should output something like this:
RTEMS Source Builder - Set Builder, 5.1.0
Build Set: 5/rtems-libbsd
config: tools/rtems-libbsd-5.cfg
package: rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1
building: rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1
sizes: rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1: 1.199GB (installed: 116.541MB)
cleaning: rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1
reporting: tools/rtems-libbsd-5.cfg -> rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1.txt
reporting: tools/rtems-libbsd-5.cfg -> rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1.xml
installing: rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1 -> $HOME/quick-start/rtems/5
cleaning: rtems-libbsd-v3cc039cdac77272a8e16b33ae5a53ccd89edf989-sparc-rtems5-1
Build Set: Time 0:00:51.898231
Note
Not all packages will build or run with all BSPs. Please ask on the Users Mailing List if you have any issues.
2.10. GSoC Getting Started¶
The goal of this page is to help new users, especially students get RTEMS compiled and running so they can start with the real work.
Please join the Users Mailing List and Developers Mailing List and ask questions. Help correct any deficiencies in the code or documentation you spot, including those on the wiki. The ultimate goal of GSoC is to help you become part of the open source community.
This section will help you to quickly setup a development environment without delving into the details. For more information you can go through the other subsections under Quick Start chapter or ask on the Developers Mailing List.
You will be best served by using a GNU/Linux environment, which could be in a virtual machine, for example that uses Virtualbox and should run on most modern desktop systems. You should also be able to work with a MacOS or Windows system, but might encounter more difficulty than a *nix environment.
Setting up a development environment consists of the following steps.
Installing dependencies for your host operating system.
Choosing an installation prefix.
Downloading the source code.
Installing the tool suite.
Building the Board Support Package (BSP).
Testing the Board Support Package (BSP).
2.10.1. Installing Dependencies¶
You need tools for your host’s operating system to build the RTEMS tool suite from source. Please have a look at the Host Computer chapter for the instructions to install the tools for your OS.
2.10.2. Choosing an installation prefix¶
The term prefix refers to the path on your computer where the software is to be installed. You can refer to the Prefix section for details on choosing an installation prefix.
2.10.3. Downloading the Sources¶
We will be using Git to clone the sources for RTEMS and RSB. This is the preferred way if you are planning to make contributions to the RTEMS project.
Please refer to the Git section for instructions on obtaining sources using Git.
2.10.4. Installing the Tool Suite¶
The Tools suite is the collection of tools required to build the BSP. This includes the compiler, debugger, assembler and other tools. These tools are architecture-specific. We will be installing the SPARC tool suite since we are building a SPARC based BSP.
Please refer to the Install the Tool Suite section for instructions on building and installing the tool suite.
2.10.5. Building the Board Support Package¶
There are two ways of building a BSP. We could either ask RSB to build the BSP or manually build it. In this section will we be building it manually. Please refer the Manual BSP Build section for the instructions.
2.10.6. Testing the Board Support Package¶
Testing is an essential part of RTEMS development process. The main reason for choosing the SPARC erc32 BSP is that, it has very good simulator support. This will allow you to test your changes without the need for SPARC hardware.
Please refer to Test a Board Support Package (BSP) for instructions on testing the BSP.
2.10.7. Prove You Can Work On RTEMS¶
This section is only for students interested in Google Summer of Code.
You have to finish the following task to prove that you can work on RTEMS.
Modify the hello world example to include a new different print statement. Something like “Hello from The Dark Side!”. Then send us enough to prove to us that you did this. We want to know you can work with RTEMS.
Create a patch of your changes and send it to Developers Mailing List along with the screenshot of the output.
If you followed this guide, this hello world modification will likely need to be made in $HOME/quick-start/src/rtems/testsuites/samples/hello/init.c. To test your changes, you have to build the BSP again. This could be done by running make in the BSP build directory.
cd $HOME/quick-start/build/b-erc32
make
If you are happy with your changes you can commit the changes and send the patch to Developers Mailing List.
2.10.8. Creating and Sending Patches¶
Before sending patches, make sure that the changes you have made conforms to RTEMS coding standards. You can refer to Contributing section for instruction on creating and sending patches.
Here are a few pointers to keep in mind while creating the patches.
Make sure not to commit changes in the master branch. This is to avoid merge conflicts when you are pulling the latest changes from the remote branch.
Avoid trailing whitespace errors.
The author name of the patch is your full name.
The author email of the patch is your valid email address.
Ensure that your patches build before sending them for review.
3. Support and Contributing¶
3.1. RTEMS Project Support¶
3.1.1. Users Mailing List¶
RTEMS offers a variety of support options and ways to contribute to the project. Users can ask their questions on the Users Mailing List. This is a low frequency mailing list intended for topics related to the use of RTEMS. If you are new to RTEMS, please join the list and ask whatever you want.
3.1.2. Documentation¶
You find the latest set of manuals at the Documentation Site.
3.1.3. All Mailing Lists¶
We have several mailing lists for RTEMS users and developers:
Announce Mailing List: Announcements for major and other project-related issues.
Bugs Mailing List: Emails generated by the Bugs Database.
Developers Mailing List: For developers of RTEMS itself.
Build Logs: Results from building and testing of RTEMS.
Users Mailing List: For users of RTEMS.
Version Control Mailing List: Commits to the RTEMS Project repositories.
3.1.4. IRC¶
RTEMS IRC is available on the Freenode network. See the Freenode web site for details on connecting, selecting a nickname, and general usage tips. If you are new to IRC it is recommended reading.
These is currently only one IRC channel available for RTEMS:
#rtems
This is a general channel for all things RTEMS. You can just hang out with other RTEMS users and developers to talk about RTEMS, using RTEMS or to make contact with other RTEMS users.
The #rtems channel is logged. You can find the logs at:
You can search the logs using Google by adding
site:rtems.org inurl:irclogs
to your search terms.
3.2. Report Bugs¶
The RTEMS Project uses a ticket system to deal with bugs, organize enhancement requests, and manage small tasks and projects. You can submit a bug report to the RTEMS Project ticket system. Before you do this, please read the following information. Good bug reports are more likely to get addressed quickly. If you have patches not specifically related to bugs or existing tickets, please have a look at the Contributing guidelines.
3.2.1. Search for Existing Bugs¶
You can search for existing bugs in the RTEMS Project ticket system. Please try to avoid duplicate bug reports and search for an existing bug before you report a new bug. If you are unsure, please ask on the Users Mailing List and we will help you sort it out.
3.2.2. Not RTEMS Bugs¶
Some issues appear to be an RTEMS bug to you, but are actually the intended behaviour or in the scope of other projects.
Bugs in the assembler, the linker, or the C library (Newlib) are not RTEMS bugs. These are separate projects, with separate mailing lists and different bug reporting procedures. The RTEMS Project is happy to work with you and those projects to resolve the problem but we must work with those projects. Bugs in those products must be addressed in the corresponding project. Report assembler, linker, and GDB bugs to sourceware.org, compiler bugs to GCC, and Newlib bugs to the Newlib mailing list. If the bug was fixed, then you can update the Source Builder to pick up the fix.
Questions about the correctness or the expected behaviour of programming language constructs or calls to library routines that are not part of RTEMS belong somewhere else.
The POSIX standard does not specify the default set of thread attributes. Thus, when passing a NULL for attributes to pthread_create(), the application is not guaranteed any particular thread behaviour.
The defaults for all RTEMS Application Configuration parameters are intentionally small. Thus, it is common for RTEMS tasking and file related calls to return errors indicating out of resources until the configuration parameters are properly tuned for the application. For example, there are only three file descriptors available by default: stdin, stdout, and stderr. Any attempt to open a socket or file will fail unless more file descriptors are configured.
When first developing a BSP, many users encounter an unexpected interrupt or exception immediately upon completion of RTEMS initialization. This occurs because interrupts are disabled during RTEMS initialization and are automatically initialized as part of switching to the first task. The interrupted user code will be in either _CPU_Context_switch() or _Thread_Handler(). This indicates that an interrupt source has not been properly initialized or masked.
Some users encounter a random reset during BSP initialization. This usually indicates that the board has a watchdog timer that is not being properly serviced during the BSP initialization.
Bugs in releases or snapshots of RTEMS not issued by the RTEMS Project. Report them to whoever provided you with the release.
3.2.3. Good Bug Reports¶
Please open the page to submit a bug to the RTEMS Project ticket system and follow the guidelines below to write a good bug report.
Provide a useful single line Summary.
Use WikiFormatting to structure the information you provide. It does help the readability of the information you provide.
Add a description of the expected behaviour. The expected behaviour may be obvious to you, but maybe not to someone else reading the bug report.
Add a description of the actual undesired behaviour.
Name the target hardware (processor architecture, chip family or model, and BSP) in the description.
Add the toolchain version used (GCC, Binutils, Newlib) to the description. Custom toolchain builds are discouraged. To avoid problems caused by custom builds of the toolchain, please build your toolchain with the Source Builder. If you use a custom build of the toolchain, then try to reproduce the bug first using a toolchain built by the RSB.
Provide the configuration options used to build the RTEMS BSP in the description. This helps to reproduce the issue.
Make the bug reproducible by others. Write a self-contained piece of source code which can be compiled and reproduces the bug. Avoid adding assembly files (*.s) produced by the compiler, or any binary files, such as object files, executables, core files, or precompiled header files. If it is difficult or time consuming to reproduce the bug, then it may not get the attention it deserves from others. Developing and debugging real-time embedded systems can be difficult. Exercise caution in reporting an error that occurs only some of the times a certain program is executed, such that retrying a sufficient number of times results in a successful compilation; this is often a symptom of a hardware problem or application issue, not of a RTEMS bug (sorry). We do recognise that sometimes a timing bug will exist in RTEMS, but we want you to exercise due diligence before pointing fingers.
Only when your bug report requires multiple source files to be reproduced should you attach an archive. Otherwise, the uploaded individual source file or diff should contain the minimal source code needed to reproduce the bug. In any case, make sure the above are included in the body of your bug report as plain text, even if needlessly duplicated as part of an archive.
Please try to reproduce the bug on the current Git master. If it is not reproducible on the Git master, you should figure out if the bug was already fixed. You can search the existing bugs once again, ask on the Users Mailing List, or do a Git bisect to find a commit which fixed the bug.
Include only information relevant to the bug.
Write separate bug reports for different bugs.
Select a Type for the ticket.
Use
defectfor a bug.Use
enhancementfor a feature request in the software or an addition to the documentation.Note
infrais used to report issues with the RTEMS servers at OSUOSL.
Select a Version for the ticket. This should be the first RTEMS version which is affected by this bug. If this is the current Git master branch use the version of the next release. Please provide the exact version of RTEMS in the description. If you use an RTEMS release, then the release number. If you use a Git clone, then the commit hash. The commit hash should be present in an RTEMS Project repository. Commit hashes of private branches are not interesting.
Select a Component for the ticket. Use
unspecifiedif you are unsure.Select a Severity for the ticket.
The fields Milestone and Priority will be most likely set by an RTEMS maintainer.
You can relate your new bug to existing bugs through the Blocked by and Blocking fields.
If you have any external files, such as screenshots or examples, please attach these as files to the ticket. Do not use external hosting because if you do use external hosting, then our historical record is broken when those files are no longer available.
Some fields should only be set by the maintainers, as it is not always clear what they should be set to. Feel free to make your own choices.
When you have checked that your report meets the criteria for a good bug report, please click on the Create ticket button to submit it to the RTEMS Project ticket system.
If you fail to supply enough information for a bug report to be reproduced, someone will probably ask you to post additional information. In this case, please post the additional information and not just to the person who requested it, unless explicitly told so.
3.2.4. Nobody Fixes my Bug¶
Sometimes, you may notice that after some time your bug report gets no attention and the bug is not magically fixed. This may have several reasons
the bug report is incomplete or confusing,
the target hardware is not available to others,
the bug is not reproducible on the Git master,
the bug is not reproducible at all,
the RTEMS version is quite old and no longer used by RTEMS maintainers, or
fixing the bug has a low priority for others.
Please note that you do not have a service contract with the RTEMS Project. The RTEMS Project is run by volunteers and persons who take care about how RTEMS performs in their application domain. If your bug does not affect the interest of someone else, then you should try to fix the bug on your own, see the Contributing guidelines. To change the priorities of others with respect to your bug, you may refer to the Commercial Support Services.
3.3. Contributing¶
3.3.1. How to Contribute?¶
You can contribute to the RTEMS Project in various ways, for example:
participation in mailing list discussions, helping other users
documentation updates, clarifications, consolidation, fixes
bug fixes, bug report consolidation
new BSPs
new device drivers
new CPU (processor architecture) ports
improvements in the existing code base (code size, code clarity, test coverage, performance optimizations)
new features
RTEMS Tools improvements
Most contributions will end up in patches of the RTEMS source code or documentation sources. The patch integration into the RTEMS repositories is done through a patch review process on the Developers Mailing List.
3.3.2. Preparing and Sending Patches¶
The RTEMS Project uses Git for version control. Git has a special command to prepare patches intended for mailing lists: git format-patch. Create logically connected patches as a patch series ideally accompanied by a cover letter (--cover-letter option). You can send patches via email through a Git command: git send-email.
3.3.3. Checklist for Patches¶
Check the following items before you send a patch to the Developers Mailing List:
The author name of the patch is your full name.
The author email of the patch is your valid email address.
The licence conditions of the contributed content allow an integration into the RTEMS code base.
If you are the copyright holder of the entire patch content, then please contribute it under the BSD-2-Clause license. For documentation use CC BY-SA 4.0.
Make sure you have a pregnant subject which does not exceed 50 characters in one line. Use a “topic: The pregnant subject” style. A topic could be the main component of patch. Just have a look at existing commit messages.
The patch has a good commit message. It should describe the reason for the change. It should list alternative approaches and why they were not chosen.
The code changes honour the coding style. At least do your changes in the style of the surrounding code.
The patch contains no spelling mistakes and grammar errors.
The patch is easy to review. It changes one thing only and contains no unrelated changes. Format changes should be separated from functional changes.
If the patch corresponds to a ticket, it should have “Close #X.” or “Update #X.” as the last line in the commit message to update status once it is committed to the repository.
The patch builds. All RTEMS tests link with this patch.
The patch does not introduce new compiler warnings.
The patch does not introduce new test failures in existing tests.
3.3.4. Patch Review Process¶
Patches sent to the Developers Mailing List undergo a patch review process. Once a patch series is accepted for integration into the RTEMS code base it is committed by an RTEMS maintainer. The maintainers are usually quite busy with all sorts of stuff. If you do not get a response to a patch series submission to the mailing list after five work days, please send a reminder. It helps if you follow the Checklist for Patches. An easy to review patch series which meets the quality standards of the RTEMS Project will be more likely get integrated quickly.
RTEMS Patch Review Process.¶
3.3.5. Why Contribute?¶
If you are writing a major extension to RTEMS, such as a port to a new CPU family (processor architecture) or model, a new target board, a major rewrite of some existing component, or adding some missing functionality, please keep in mind the importance of keeping other developers informed. Part of being a good cooperating member of the RTEMS development team is the responsibility to consider what the other developers need in order to work effectively.
Nobody likes to do a lot of work and find it was duplicated effort. So when you work on a major new feature, you should tell Users Mailing List what you are working on, and give occasional reports of how far you have come and how confident you are that you will finish the job. This way, other developers (if they are paying attention) will be aware which projects would duplicate your effort, and can either join up with you, or at least avoid spending time on something that will be unnecessary because of your work. If, for whatever reason, you are not in a position to publicly discuss your work, please at least privately let an RTEMS maintainer know about it so they can look out for duplicated effort or possible collaborators.
You should also monitor the Users Mailing List and Developers Mailing List to see if someone else mentions working on a similar project to yours. If that happens, speak up!
If you are thinking of taking a contract to develop changes under a temporary delayed-release agreement, please negotiate the agreement so that you can give progress reports before the release date, even though you cannot release the code itself. Also please arrange so that, when the agreed-on date comes, you can release whatever part of the job you succeeded in doing, even if you have not succeeded in finishing it. Someone else may be able to finish the job.
Many people have done RTEMS ports or BSPs on their own, to a wide variety of processors, without much communication with the RTEMS development team. However, much of this work has been lost over time, or have proven very hard to integrate. So, what we are asking is that, to the maximum extent possible, you communicate with us as early on and as much as possible.
3.3.6. Common Questions and Answers¶
Here are some questions RTEMS porters may have with our answers to them. While the focus here is on new ports and BSPs, we believe that the issues are similar for other RTEMS development efforts including student efforts to implement new algorithmic optimizations.
Our engineers understand our target environment better than anyone else, and we have a tight schedule. Why should we work with the RTEMS developers, when we can get the code out faster by whacking it out on our own?
You understand your target environment better than anyone else. However, the RTEMS developers understand RTEMS better than anyone else; furthermore, the RTEMS developers tend to have a wide breadth of experience across a large number of processors, boards, peripherals, and application domains. It has been our experience that few problems encountered in embedded systems development are unique to a particular processor or application. The vast majority of the time an issue that arises in one project has also shown up in other projects.
The intimate knowledge of RTEMS internals as well as a wide breadth of embedded systems knowledge means that there is a good chance that at least one RTEMS developer has already addressed issues you are likely to face when doing your port, BSP, or application. The developers can help guide you towards a workable long term solution, possibly saving you significant time in your development cycle.
If getting the sources into the official RTEMS distributions is one of your goals, then engaging other RTEMS developers early will also likely shorten your development time. By interacting as early as possible you are more likely to write code which can be easily accepted into the official sources when you are finished. If you wait until you think you are done to begin interacting with the RTEMS team, you might find that you did some things wrong and you may have to rewrite parts of your RTEMS port, which is a waste of your valuable time.
Why should we care if our port is integrated into the official RTEMS sources? We can distribute it ourselves to whoever is interested.
Yes, the RTEMS licenses allows you to do that. But by doing so, you end up having to maintain that code yourself; this can be a significant effort over time as the RTEMS sources change rapidly.
You also lose the advantage of wider exposure by including your port in the official RTEMS sources maintained by the RTEMS Project. The wider exposure in the RTEMS developer and tester community will help keep your work up to date with the current sources. You may even find that volunteers will run the ever-growing test suite on your port and fix problems during the development cycle – sometimes without your intervention.
It has been our experience that integrated ports tend to ultimately be of better quality and stay up to date from release to release.
Why should we communicate up front? We are happy to let the RTEMS developers integrate our stuff later.
See above. It will save work for you over both the short and the long term, and it is the right thing to do.
Aspects of my target environment that my application exploits are still under NDA.
Nevertheless, if the target hardware is built of any commercial parts that are generally available including, but not limited to, the CPU or peripherals, then that portion of your work is still of general use. Similarly, if you have written software that adheres to existing API or interface standards, then that portion is also of general use. Our experience is that most embedded applications do utilize a custom mix of hardware and application, but they are built upon layers of hardware and software components that are in no way unique to the project.
If you are porting to an unreleased CPU family or model, then just announcing it is important because other RTEMS users may be planning to use it and some of them may already be trying to port RTEMS on their own. Your customers might be happier to know that your port will eventually be available. Also, there is no requirement that RTEMS include all features or ports at any particular time, so you are encouraged to submit discrete pieces of functionality in stages.
Assume that your processor has some new functionality or peripherals. However that functionality is still covered by NDA, but the basic core architecture is not. It is still to your advantage to go ahead and work with the developers early to provide a “base port” for the CPU family. That base port would only use the publicly available specifications until such time as the NDA is lifted. Once the NDA is lifted you can work with the developers to provide the code necessary to take advantage of the new functionality.
Ultimately, cooperating with the free software community as early as possible helps you by decreasing your development cycle, decreasing your long term maintenance costs and may help raise interest in your processor by having a free compiler implementation available to anyone who wants to take a look.
3.4. Commercial Support Services¶
The wider RTEMS community has developers and organizations who can provide commercial support services. These services range from training, implementing new features in RTEMS, deployment of RTEMS, helping establish a new project environment for a team, to application and system design.
The RTEMS Project does not endorse or promote any provider of these services and we recommend you use a search engine to locate a suitable provider. If you are unsure please contact a provider and see what is available.
If you develop a new feature or you have someone do this for you we recommend you have the work submitted to the project and merged. Once accepted into the project the work will be maintained as part of the development process within the project and this is a benefit for.
4. Host Computer¶
RTEMS applications are developed using cross-development tools running on a development computer, more often called the host computer. These are typically your desktop machine or a special build server. All RTEMS tools and runtime libraries are built from source on your host machine. The RTEMS Project does not maintain binary builds of the tools. This differs to what you normally experience with host operating systems, and it is, however this approach works well. RTEMS is not a host operating system and it is not a distrbution. Deploying binary packages for every possible host operating system is too big a task for the RTEMS Project and it is not a good use of core developer time. Their time is better spent making RTEMS better and faster.
The RTEMS Project’s aim is to give you complete freedom to decide on the languages used in your project, which version control system, and the build system for your application.
The rule for selecting a computer for a developer is more is better but we do understand there are limits. Projects set up different configurations, some have a development machine per developer while others set up a tightly controlled central build server. RTEMS Ecosystem is flexible and lets you engineer a development environment that suites you. The basic specs are:
Multicore processor
8G bytes RAM
256G harddisk
RTEMS makes no demands on graphics.
If you are using a VM or your host computer is not a fast modern machine do not be concerned. The tools may take longer to build than faster hardware however building tools is something you do once. Once the tools and RTEMS is built all your time can be spent writing and developing your application. Over an hour can happen and for the ARM architecture and with all BSPs it can be many hours.
4.1. Host Operating Systems¶
A wide range of host operating systems and hardware can be used. The host operating systems supported are:
Linux
FreeBSD
NetBSD
Apple OS X
Windows
Solaris
The functionality on a POSIX operating such as Linux and FreeBSD is similar and most features on Windows are supported but you are best to ask on the Users Mailing List if you have a specific question.
We recommend you maintain your operating system by installing any updates.
We also recommend you keep your environment to the bare minimum, particularly the PATH variable. Using environment variables has been proven over the years to be difficult to manage in production systems.
Warning
The RSB assumes your host is set up and the needed packages are installed and configured to work. If your host has not been set up please refer to section that covers your host’s packages you need to install.
Path to use when building applications:
Do not forget to set the path before you use the tools, for example to build the RTEMS kernel.
The RSB by default will install (copy) the executables to a directory tree under the prefix you supply. To use the tools once finished just set your path to the bin directory under the prefix you use. In the examples that follow the prefix is $HOME/development/rtems/4.11 and is set using the --prefix option so the path you need to configure to build applications can be set with the following in a BASH shell:
$ export PATH=$HOME/development/rtems/4.11/bin:$PATH
Make sure you place the RTEMS tool path at the front of your path so they are searched first. RTEMS can provide newer versions of some tools your operating system provides and placing the RTEMS tools path at the front means it is searched first and the RTEMS needed versions of the tools are used.
Warning
Do not put spaces or special characters in the directories you use to build RTEMS. Many of the packages built by the RSB use GNU make, which cannot handle spaces in pathnames. If there is a space in the pathname the build will fail. Special characters are also likely to confuse build systems.
Note
RSB and RTEMS have a matching git branch for each version of RTEMS. For example, if you want to build a toolchain for 4.11, then you should checkout the 4.11 branch of the RSB:
$ git checkout -t origin/4.11
Branches are available for the 4.9, 4.10, and 4.11 versions of RTEMS.
4.2. POSIX Hosts¶
POSIX hosts are most Unix operating systems such as Linux, FreeBSD and NetBSD. RTEMS development works well on Unix and can scale from a single user and a desktop machine to a team with decentralised or centralised development infrastructure.
4.2.1. Root Access¶
You either have root access to your host development machine or you do not. Some users are given hardware that is centrally managed. If you do not have root access you can create your work environment in your home directory. You could use a prefix of $HOME/development/rtems or $HOME/rtems. Note, the $HOME environment variable can be substituted with ~.
Choose an Installation Prefix details using Prefixes to manage the installation.
RTEMS Tools and packages do not require root access to be built and we encourage you to not build the tools as root. If you need to control write access then it is best to manage this with groups assigned to users.
If you have root access you can decide to install the tools under any suitable prefix. This may depend on the hardware in your host development machine. If the machine is a centralised build server the prefix may be used to separate production versions from the test versions and the prefix paths may have restricted access rights to only those who manage and have configuration control of the machine. We call this project sandboxing and Project Sandboxing explains this in more detail.
4.2.2. Linux¶
BSP Build will require pax package if RTEMS is configured with the --enable-tests option, see Building RTEMS Tests. This package is not installed , by default, on many Linux distributions, you can check for it using your package manager. Install it, if it is not present on your system.
A number of different Linux distrubutions are known to work. The following have been tested and report as working.
4.2.2.1. ArchLinux¶
The following packages are required on a fresh Archlinux 64bit installation:
# pacman -S base-devel gdb xz unzip ncurses git zlib
Archlinux, by default installs texinfo-5 which is incompatible for building GCC 4.7 tree. You will have to obtain texinfo-legacy from AUR and provide a manual override:
# pacman -R texinfo
$ yaourt -S texinfo-legacy
# ln -s /usr/bin/makeinfo-4.13a /usr/bin/makeinfo
4.2.2.2. CentOS¶
The following packages are required on a minimal CentOS 6.3 or Cent)S 7 64-bit installation:
# yum install autoconf automake binutils gcc gcc-c++ gdb make patch pax \
bison flex xz unzip ncurses-devel texinfo zlib-devel python-devel git
On CentOS 8, the pax command is now provided by the spax package, you need to enable the PowerTools repository. and use Python3. On a fresh install, the following commands should install everything you need for RTEMS development:
# dnf install yum-utils
# dnf config-manager --set-enabled PowerTools
# dnf update
# dnf groupinstall "Development Tools"
# dnf install python3 python3-pip python3-setuptools python3-devel
# dnf install texinfo spax
# alternatives --set python /usr/bin/python3
The minimal CentOS distribution is a specific DVD that installs a minimal system. If you use a full system some of these packages may have been installed.
4.2.2.3. Fedora¶
The RTEMS Source Builder has been tested on Fedora 19 64bit with the following packages:
# yum install ncurses-devel python-devel git bison gcc cvs gcc-c++ \
flex texinfo patch perl-Text-ParseWords zlib-devel
4.2.2.4. Raspbian¶
The is the Debian distribution for the Raspberry Pi. The following packages are required:
$ sudo apt-get install autoconf automake bison flex binutils gcc g++ gdb \
texinfo unzip ncurses-dev python-dev git
It is recommended you get Model B of the Pi with 512M of memory and to mount a remote disk over the network. The tools can be built on the network disk with a prefix under your home directory as recommended and end up on the SD card.
4.2.2.5. Ubuntu¶
The latest version is Ubuntu 18.04.1 LTS 64-bit. This section also includes Xubuntu. A minimal installation was used and the following packages installed:
$ sudo apt-get build-dep build-essential gcc-defaults g++ gdb git \
unzip pax bison flex texinfo unzip python3-dev libpython-dev \
libncurses5-dev zlib1g-dev
Note that in previous versions of Ubuntu, the package libpython-dev was python2.7-dev. The name of packages changes over time. You need the package with Python development libraries for C/C++ programs. The following is needed for recent versions:
$ sudo apt-get install python-dev
It is likely necessary that you will have to enable the Ubuntu Source Repositories. Users have suggested the following web pages which have instructions:
4.2.2.6. Linux Mint¶
zlib package is required on Linux Mint. It has a different name (other than the usual zlib-dev):
# sudo apt-get install zlib1g-dev
4.2.2.7. openSUSE¶
This has been reported to work but no instructions were provided. This is an opportunity to contribute. Please submit any guidance you can provide.
4.2.3. FreeBSD¶
The RTEMS Source Builder has been tested on FreeBSD 9.1, 10.3, 11 and 12 64bit versions. You need to install some ports. They are:
# pkg install -y python
# pkg install -y gsed
FreeBSD’s default C compiler is LLVM and installing the host’s GCC compiler package may break building GCC. We recommend you do not install the GCC package and you use the default C compiler.
If you wish to build Windows (mingw32) tools please install the following ports:
# pkg install -y mingw32-binutils mingw32-gcc
# pkg install -y mingw32-zlib mingw32-pthreads
The zlip and pthreads ports for MinGW32 are used when builiding a Windows QEMU.
Check if your kernel has a /dev/fd directory. If it does not we recommend you run as root the following command to speed up Python 3’s subprocess support:
# mount -t fdescfs none /dev/fd
The support speeds up closing file descriptors when creating subprocesses.
4.2.4. NetBSD¶
The RTEMS Source Builder has been tested on NetBSD 6.1 i386. Packages to add are:
# pkg_add ftp://ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/i386/6.1/devel/gmake-3.82nb7.tgz
# pkg_add ftp://ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/i386/6.1/devel/bison-2.7.1.tgz
# pkg_add ftp://ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/i386/6.1/archivers/xz-5.0.4.tgz
4.3. Apple macOS¶
Apple’s macOS is fully supported. You need to download and install a recent version of the Apple developer application Xcode. Xocde is available in the App Store. Make sure you install the Command Line Tools add on available for download within Xcode and once installed open a Terminal shell and enter the command cc and accept the license agreement.
The normal prefix when working on macOS as a user is under your home directory. Prefixes of $HOME/development/rtems or $HOME/rtems are suitable.
Choose an Installation Prefix details using Prefixes to manage the installation.
4.3.1. Catalina¶
In the macOS Catalina 10.15 Release Notes Apple deprecated several scripting language runtimes such as Python 2.7. See also Xcode 11 Release Notes. Due to the deprecated Python 2.7 support, we recommend to install and use the latest Python 3 release from python.org.
4.3.2. Sierra¶
The RSB works on Sierra with the latest Xcode.
4.3.3. Mavericks¶
The RSB works on Mavericks and the GNU tools can be built for RTEMS using the Mavericks clang LLVM tool chain. You will need to build and install a couple of packages to make the RSB pass the sb-check. These are CVS and XZ. You can get these tools from a packaging tool for macOS such as MacPorts or HomeBrew.
I do not use third-party packaging on macOS and prefer to build the packages from source using a prefix of /usr/local. There are good third-party packages around however they sometimes bring in extra dependence and that complicates my build environment and I want to know the minimal requirements when building tools. The following are required:
- . The XZ package’s home page is http://tukaani.org/xz/ and I use version
5.0.5. XZ builds and installs cleanly.
4.4. Microsoft Windows¶
RTEMS supports Windows as a development host and the tools for most architectures are available. The RTEMS Project relies on the GNU tools for compilers and debuggers and we use the simulators that come with GDB and QEMU. The Windows support for these tools varies and the RTEMS Project is committed to helping the open source community improve the Windows experience. If something is not working or supported please email the Users Mailing List.
The RTEMS Project’s Windows tools can be native Windows executables which give the user the best possible experience on Windows. Native Windows programs use the standard Windows DLLs and paths. Integration with standard Windows integrated development tools such as editors is straight forward. POSIX emulation environments such as Cygwin and the MSYS2 shell have special executables that require a POSIX emulation DLL and these emulation DLLs add an extra layer of complexity as well as a performance over-head. The RTEMS Project uses these POSIX emulation shells to run configure scripts that come with various open source packages such as gcc so they form an important and valued part of the environment we describe here. The output of this procedure forms the tools you use during your application development and they do not depend on the emulation DLLs.
The performance of a native Windows compiler is as good as you can have on Windows and the performance compiling a single file will be similar to that on a host like Linux or FreeBSD given the same hardware. Building the tools from source is much slower on Windows because POSIX shells and related tools are used and the POSIX emulation overhead it much much slower than a native POSIX operating system like Linux and FreeBSD. This overhead is only during the building of the tools and the RTEMS kernel and if you use a suitable build system that is native to Windows your application development should be similar to other operating systems.
Building is known to work on Windows 7 64bit Professional and Windows 10 64bit.
4.4.1. Windows Path Length¶
Windows path length is limited and can cause problems when building the tools. The standard Windows API has a MAX_PATH length of 260 characters. This can effect some of the tools used by RTEMS. It is recommended you keep the top level directories as short as possible when building the RTEMS tools and you should also keep an eye on the path length when developing your application. The RTEMS built tools can handle much longer path lengths however some of the GNU tools such as those in the binutils package cannot.
The release packages of the RSB when unpacked have top level file names that are too big to build RTEMS. You need to change or rename that path to something smaller to build. This is indicated in Releases.
4.4.2. Windows Spaces In Paths¶
Occasionally, a program will fail on Windows with errors that appear as if a directory or file name was partially parsed by some utility or program. This can be caused by having directories of file names with spaces. Programs written in scripting languages sometimes fail to properly quote file names and the space is incorrectly interpreted.
Parts of the PATH inherited from the native Windows environment often include directory names with spaces. Sometimes it is necessary to set the PATH explicitly to avoid these.
4.4.3. Parallel Builds with Make¶
The MSYS2 GNU make has problems when using the jobs option. The RSB defaults to automatically using as many cores as the host machine has. To get a successful build on Windows it is recommended you add the --jobs=none option to all RSB build set commands.
4.4.4. POSIX Support¶
Building the RTEMS compilers, debugger, the RTEMS kernel and a number of other third-party packages requires a POSIX environment. On Windows you can use Cygwin or MSYS2. This document focuses on MSYS2. It is smaller than Cygwin and comes with the Arch Linux package manager pacman.
MSYS2 provides MinGW64 support as well as a POSIX shell called MSYS2. The MinGW64 compiler and related tools produce 64bit native Windows executables. The shell is a standard Bourne shell and the MSYS2 environment is a stripped Cygwin shell with enough support to run the various configure scripts needed to build the RTEMS tools and the RTEMS kernel.
MSYS2 is built around the pacman packaging tool. This makes MSYS2 a distribution and that is a welcome feature on Windows. You get a powerful tool to manage your development environment on Windows.
4.4.5. Python¶
We need Python to build the tools as the RSB is written in Python and we need suitable Python libraries to link to GDB as RTEMS makes use of GDB’s Python support. This places specific demands on the Python we need installed and available and MSYS2 provides suitable Python versions we can use. You need to make sure you have the correct type and version of Python installed.
We cannot use the Python executables created by the Python project (python.org) as they are built by Microsoft’s C (MSC) compiler. Linking the MSC Python libraries with the MinGW64 executables is not easy and MSYS provides us with a simple solution so we do not support linking MSC libraries.
MSYS2 provides two types and two versions of Python executables, MinGW and MSYS and Python version 2 and 3. For Windows we need the MinGW executable so we have suitables libraries and we have to have Python version 2 because on Windows GDB only builds with Python2.
You also need to install the MSYS version of Python along with the MinGW64 Python2 package. The MSYS Python is version 3 and the RSB can support version 2 and 3 of Python and it helps handle some of the long paths building GCC can generate.
4.4.6. MSYS2¶
MSYS2 is installed on a new machine using the MSYS2 installer found on https://msys2.github.io/. Please select the x86_64 variant for 64bit support. Run the installer following the 7 steps listed on the page.
MSYS2 uses the pacman package manager. The Arch Linux project has detailed documentation on how to use pacman. What is shown here is a just few examples of what you can do.
Open a 64bit MSYS shell from the Start Menu:
The packages we require are:
python
mingw-w64-x86_64-python2
mingw-w64-x86_64-gcc
git
bison
cvs
diffutils
make
patch
tar
texinfo
unzip
Note
The actual output provided may vary due to changes in the dependent packages or newer package versions.
Install the packages using pacman:
$ pacman -S python mingw-w64-x86_64-python2 mingw-w64-x86_64-gcc \
bison cvs diffutils git make patch tar texinfo unzip
resolving dependencies...
looking for conflicting packages...
.... output shortened for brevity ....
4.4.7. Cygwin¶
Building on Windows is a little more complicated because the Cygwin shell is used rather than the MSYS2 shell. The MSYS2 shell is simpler because the detected host triple is MinGW so the build is a standard cross-compiler build. A Canadian cross-build using Cygwin is supported if you would like native tools or you can use a Cygwin built set of tools.
Install a recent Cygwin version using the Cygwin setup tool. Select and install the groups and packages listed:
Group | Package |
Archive | bsdtar |
Archive | unzip |
Archive | xz |
Devel | autoconf |
Devel | autoconf2.1 |
Devel | autoconf2.5 |
Devel | automake |
Devel | binutils |
Devel | bison |
Devel | flex |
Devel | gcc4-core |
Devel | gcc4-g++ |
Devel | git |
Devel | make |
Devel | mingw64-x86_64-binutils |
Devel | mingw64-x86_64-gcc-core |
Devel | mingw64-x86_64-g++ |
Devel | mingw64-x86_64-runtime |
Devel | mingw64-x86_64-zlib |
Devel | patch |
Devel | zlib-devel |
MinGW | mingw-zlib-devel |
Python | python |
The setup tool will add a number of dependent package and it is ok to accept them.
Disabling Windows Defender improves performance if you have another up to date virus detection tool installed and enabled. The excellent Process Hacker 2 tool can monitor the performance and the Windows Defender service contributed a high load. In this case a third-party virus tool was installed so the Windows Defender service was not needed.
To build a MinGW tool chain a Canadian cross-compile (Cxc) is required on Cygwin because the host is Cygwin therefore a traditional cross-compile will result in Cygiwn binaries. With a Canadian cross-compile a Cygwin cross-compiler is built as well as the MinGW RTEMS cross-compiler. The Cygwin cross-compiler is required to build the C runtime for the RTEMS target because we are building under Cygiwn. The build output for an RTEMS 4.10 ARM tool set is:
chris@cygwin ~/development/rtems/src/rtems-source-builder/rtems
$ ../source-builder/sb-set-builder --log=l-arm.txt \
--prefix=$HOME/development/rtems/4.10 4.10/rtems-arm
RTEMS Source Builder - Set Builder, v0.2
Build Set: 4.10/rtems-arm
config: expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-w64-mingw32-1
building: expat-2.1.0-x86_64-w64-mingw32-1
reporting: expat-2.1.0-1.cfg -> expat-2.1.0-x86_64-w64-mingw32-1.html
config: tools/rtems-binutils-2.20.1-1.cfg
package: arm-rtems4.10-binutils-2.20.1-1 <1>
building: arm-rtems4.10-binutils-2.20.1-1
package: (Cxc) arm-rtems4.10-binutils-2.20.1-1 <2>
building: (Cxc) arm-rtems4.10-binutils-2.20.1-1
reporting: tools/rtems-binutils-2.20.1-1.cfg ->
arm-rtems4.10-binutils-2.20.1-1.html
config: tools/rtems-gcc-4.4.7-newlib-1.18.0-1.cfg
package: arm-rtems4.10-gcc-4.4.7-newlib-1.18.0-1
building: arm-rtems4.10-gcc-4.4.7-newlib-1.18.0-1
package: (Cxc) arm-rtems4.10-gcc-4.4.7-newlib-1.18.0-1
building: (Cxc) arm-rtems4.10-gcc-4.4.7-newlib-1.18.0-1
reporting: tools/rtems-gcc-4.4.7-newlib-1.18.0-1.cfg ->
arm-rtems4.10-gcc-4.4.7-newlib-1.18.0-1.html
config: tools/rtems-gdb-7.3.1-1.cfg
package: arm-rtems4.10-gdb-7.3.1-1
building: arm-rtems4.10-gdb-7.3.1-1
reporting: tools/rtems-gdb-7.3.1-1.cfg -> arm-rtems4.10-gdb-7.3.1-1.html
config: tools/rtems-kernel-4.10.2.cfg
package: arm-rtems4.10-kernel-4.10.2-1
building: arm-rtems4.10-kernel-4.10.2-1
reporting: tools/rtems-kernel-4.10.2.cfg -> arm-rtems4.10-kernel-4.10.2-1.html
installing: expat-2.1.0-x86_64-w64-mingw32-1 -> /cygdrive/c/Users/chris/development/rtems/4.10
installing: arm-rtems4.10-binutils-2.20.1-1 -> /cygdrive/c/Users/chris/development/rtems/4.10 <3>
installing: arm-rtems4.10-gcc-4.4.7-newlib-1.18.0-1 -> /cygdrive/c/Users/chris/development/rtems/4.10
installing: arm-rtems4.10-gdb-7.3.1-1 -> /cygdrive/c/Users/chris/development/rtems/4.10
installing: arm-rtems4.10-kernel-4.10.2-1 -> /cygdrive/c/Users/chris/development/rtems/4.10
cleaning: expat-2.1.0-x86_64-w64-mingw32-1
cleaning: arm-rtems4.10-binutils-2.20.1-1
cleaning: arm-rtems4.10-gcc-4.4.7-newlib-1.18.0-1
cleaning: arm-rtems4.10-gdb-7.3.1-1
cleaning: arm-rtems4.10-kernel-4.10.2-1
Build Set: Time 10:09:42.810547 <4>
Items:
The Cygwin version of the ARM cross-binutils.
The +(Cxc)+ indicates this is the MinGW build of the package.
Only the MinGW version is installed.
Cygwin is slow so please be patient. This time was on an AMD Athlon 64bit Dual Core 6000+ running at 3GHz with 4G RAM running Windows 7 64bit.
Warning
Cygwin documents the ‘Big List Of Dodgy Apps’ or ‘BLODA’. The link is http://cygwin.com/faq/faq.html#faq.using.bloda and it is worth a look. You will see a large number of common pieces of software found on Windows systems that can cause problems. My testing has been performed with NOD32 running and I have seen some failures. The list is for all of Cygwin so I am not sure which of the listed programs effect the RTEMS Source Biulder. The following FAQ item talks about fork failures and presents some technical reasons they cannot be avoided in all cases. Cygwin and it’s fork MSYS are fantastic pieces of software in a difficult environment. I have found building a single tool tends to work, building all at once is harder.
5. Installation¶
This section details how to set up and install the RTEMS Ecosystem. You will create a set of tools and an RTEMS kernel for your selected Board Support Package (BSP).
You will be asked to follow a few simple steps and when you have finished you will have a development environment set up you can use to build applications for RTEMS. You will have also created a development environment you and a team can adapt for a project of any size and complexity.
RTEMS applications are developed using cross-development tools running on a development computer, more commonlly referred to as the host computer. These are typically your desktop machine or a special build server. All RTEMS tools and runtime libraries are built from source on your host machine. The RTEMS Project does not maintain binary builds of the tools. This may appear to be the opposite to what you normally experience with host operating systems, and it is, however this approach works well. RTEMS is not a host operating system and it is not a distrbution. Providing binary packages for every possible host operating system is too big a task for the RTEMS Project and it is not a good use of core developer time. Their time is better spent making RTEMS better and faster.
The RTEMS Project base installation set ups the tools and the RTEMS kernel for the selected BSPs. The tools run on your host computer are used to compile, link, and format executables so they can run on your target hardware.
The RTEMS Project supports two set ups, release and developer environments. Release installations create the tools and kernel in a single pass ready for you to use. The tools and kernel are stable and only bug fixes are added creating new dot point releases. The developer set up tracks the Git repositories for the tools and kernel.
5.1. Releases¶
RTEMS releases provide a stable version of the kernel for the supported architectures. RTEMS maintaines the current and previous releases. Support for older releases is provided using the RTEMS support channels.
Please read Host Computer before continuing. The following procedure assumes you have installed and configured your host operating. It also assumes you have installed any dependent packages needed when building the tools and the kernel.
You need to select a location to build and install the RTEMS Tool chain and RTEMS. Make sure there is plenty of disk space and a fast disk is recommended. Our procedure will document building and installing the tools in a base directory called /opt/rtems. This path will require root access. If you are working on a machine you do not have root access to you can use a home directory, If building on Windows use /c/opt/rtems to keep the top level paths as short as possible. Windows Path Length provides more detail about path lengths on Windows.
The location used to install the tools and kernel is called the prefix. Choose an Installation Prefix explains prefixes and how to use them. It is best to have a prefix for each different version of RTEMS you are using. If you are using RTEMS 4.11 in production it is not a good idea to install a development version of 5 over the top by using the same prefix as the 4.11 build. A separate prefix for each version avoids this.
Released versions of the RTEMS Source Builder (RSB) downloads all source code for all packages from the FTP File Server rather than from the package’s home site. Hosting all the source on the FTP File Server ensures the source is present for the life of the release on the FTP File Server. If there is a problem accessing the RTEMS FTP the RSB will fall back to the packages home site.
The FTP File Server is hosted at the Oregon State University’s The Open Source Lab (http://osuosl.org/). This is a nonprofit organization working for the advancement of open source technologies and RTEMS is very fortunate to be shosted here. It has excellent internet access and performance.
Note
Controlling the RTEMS Kernel Build
Building releases by default does not build the RTEMS kernel. To build the RTEMS kernel add the --with-rtems option to the RSB command line.
By default all the BSPs for an architecture are built. If you only wish to have a specific BSP built you can specify the BSP list by providing to the RSB the option --with-rtemsbsp. For example to build two BSPs for the SPARC architecture you can supply --with-rtemsbsp="erc32 leon3". This can speed the build time up for some architectures that have a lot of BSPs.
Once you have built the tools and kernel you can move to the Packages section of the manual.
5.1.1. RTEMS Tools and Kernel¶
This procedure will build a SPARC tool chain. Set up a suitable workspace to build the release in. On Unix:
$ cd
$ mkdir -p development/rtems/releases
$ cd development/rtems/releases
If building on Windows:
$ cd /c
$ mkdir -p opt/rtems
$ cd opt/rtems
Note the paths on Windows will be different to those shown.
Download the RTEMS Source Builder (RSB) from the RTEMS FTP server:
$ wget https://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/rtems-source-builder-4.11.0.tar.xz
--2016-03-21 10:50:04-- https://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/rtems-source-builder-4.11.0.tar.xz
Resolving ftp.rtems.org (ftp.rtems.org)... 140.211.10.151
Connecting to ftp.rtems.org (ftp.rtems.org)|140.211.10.151|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 967056 (944K) [application/x-xz]
Saving to: 'rtems-source-builder-4.11.0.tar.xz'
rtems-source-builder-4.1 100%[====================================>] 944.39K 206KB/s in 5.5s
2016-03-21 10:50:11 (173 KB/s) - 'rtems-source-builder-4.11.0.tar.xz' saved [967056/967056]
On Unix unpack the RSB release tar file using:
$ tar Jxf rtems-source-builder-4.11.0.tar.xz
$ cd rtems-source-builder-4.11.0/rtems/
On Windows you need to shorten the path (See Windows Path Length) after you have unpacked the tar file:
$ tar Jxf rtems-source-builder-4.11.0.tar.xz
$ mv rtems-source-builder-4.11.0 4.110
$ cd 4.11.0
Build a tool chain for the SPARC architecure. We are using the SPARC architecture in our example because GDB has a good simulator that lets us run and test the samples RTEMS builds by default
If building on Windows add --jobs=none to avoid GNU make issues on Windows discussed in Parallel Builds with Make.
$ ../source-builder/sb-set-builder \
--prefix=/opt/rtems/4.11 4.11/rtems-sparc
Build Set: 4.11/rtems-sparc
Build Set: 4.11/rtems-autotools.bset
Build Set: 4.11/rtems-autotools-internal.bset
config: tools/rtems-autoconf-2.69-1.cfg
package: autoconf-2.69-x86_64-freebsd10.1-1
Creating source directory: sources
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/autoconf-2.69.tar.gz -> sources/autoconf-2.69.tar.gz
downloading: sources/autoconf-2.69.tar.gz - 1.8MB of 1.8MB (100%)
building: autoconf-2.69-x86_64-freebsd10.1-1
config: tools/rtems-automake-1.12.6-1.cfg
package: automake-1.12.6-x86_64-freebsd10.1-1
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/automake-1.12.6.tar.gz -> sources/automake-1.12.6.tar.gz
downloading: sources/automake-1.12.6.tar.gz - 2.0MB of 2.0MB (100%)
Creating source directory: patches
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/automake-1.12.6-bugzilla.redhat.com-1239379.diff -> patches/automake-1.12.6-bugzilla.redhat.com-1239379.diff
downloading: patches/automake-1.12.6-bugzilla.redhat.com-1239379.diff - 408.0 bytes of 408.0 bytes (100%)
building: automake-1.12.6-x86_64-freebsd10.1-1
cleaning: autoconf-2.69-x86_64-freebsd10.1-1
cleaning: automake-1.12.6-x86_64-freebsd10.1-1
Build Set: Time 0:00:32.749337
Build Set: 4.11/rtems-autotools-base.bset
config: tools/rtems-autoconf-2.69-1.cfg
package: autoconf-2.69-x86_64-freebsd10.1-1
building: autoconf-2.69-x86_64-freebsd10.1-1
reporting: tools/rtems-autoconf-2.69-1.cfg -> autoconf-2.69-x86_64-freebsd10.1-1.txt
reporting: tools/rtems-autoconf-2.69-1.cfg -> autoconf-2.69-x86_64-freebsd10.1-1.xml
config: tools/rtems-automake-1.12.6-1.cfg
package: automake-1.12.6-x86_64-freebsd10.1-1
building: automake-1.12.6-x86_64-freebsd10.1-1
reporting: tools/rtems-automake-1.12.6-1.cfg -> automake-1.12.6-x86_64-freebsd10.1-1.txt
reporting: tools/rtems-automake-1.12.6-1.cfg -> automake-1.12.6-x86_64-freebsd10.1-1.xml
installing: autoconf-2.69-x86_64-freebsd10.1-1 -> /opt/work/rtems/4.11.0
installing: automake-1.12.6-x86_64-freebsd10.1-1 -> /opt/work/rtems/4.11.0
cleaning: autoconf-2.69-x86_64-freebsd10.1-1
cleaning: automake-1.12.6-x86_64-freebsd10.1-1
Build Set: Time 0:00:15.619219
Build Set: Time 0:00:48.371085
config: devel/expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-freebsd10.1-1
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/expat-2.1.0.tar.gz -> sources/expat-2.1.0.tar.gz
downloading: sources/expat-2.1.0.tar.gz - 549.4kB of 549.4kB (100%)
building: expat-2.1.0-x86_64-freebsd10.1-1
reporting: devel/expat-2.1.0-1.cfg -> expat-2.1.0-x86_64-freebsd10.1-1.txt
reporting: devel/expat-2.1.0-1.cfg -> expat-2.1.0-x86_64-freebsd10.1-1.xml
config: tools/rtems-binutils-2.26-1.cfg
package: sparc-rtems4.11-binutils-2.26-x86_64-freebsd10.1-1
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/binutils-2.26.tar.bz2 -> sources/binutils-2.26.tar.bz2
downloading: sources/binutils-2.26.tar.bz2 - 24.4MB of 24.4MB (100%)
building: sparc-rtems4.11-binutils-2.26-x86_64-freebsd10.1-1
reporting: tools/rtems-binutils-2.26-1.cfg ->
sparc-rtems4.11-binutils-2.26-x86_64-freebsd10.1-1.txt
reporting: tools/rtems-binutils-2.26-1.cfg ->
sparc-rtems4.11-binutils-2.26-x86_64-freebsd10.1-1.xml
config: tools/rtems-gcc-4.9.3-newlib-2.2.0-20150423-1.cfg
package: sparc-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-freebsd10.1-1
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/gcc-4.9.3.tar.bz2 -> sources/gcc-4.9.3.tar.bz2
downloading: sources/gcc-4.9.3.tar.bz2 - 85.8MB of 85.8MB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/newlib-2.2.0.20150423.tar.gz -> sources/newlib-2.2.0.20150423.tar.gz
downloading: sources/newlib-2.2.0.20150423.tar.gz - 16.7MB of 16.7MB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/mpfr-3.0.1.tar.bz2 -> sources/mpfr-3.0.1.tar.bz2
downloading: sources/mpfr-3.0.1.tar.bz2 - 1.1MB of 1.1MB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/mpc-0.8.2.tar.gz -> sources/mpc-0.8.2.tar.gz
downloading: sources/mpc-0.8.2.tar.gz - 535.5kB of 535.5kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/gmp-5.0.5.tar.bz2 -> sources/gmp-5.0.5.tar.bz2
downloading: sources/gmp-5.0.5.tar.bz2 - 2.0MB of 2.0MB (100%)
building: sparc-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-freebsd10.1-1
reporting: tools/rtems-gcc-4.9.3-newlib-2.2.0-20150423-1.cfg ->
sparc-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-freebsd10.1-1.txt
reporting: tools/rtems-gcc-4.9.3-newlib-2.2.0-20150423-1.cfg ->
sparc-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-freebsd10.1-1.xml
config: tools/rtems-gdb-7.9-1.cfg
package: sparc-rtems4.11-gdb-7.9-x86_64-freebsd10.1-1
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/gdb-7.9.tar.xz -> sources/gdb-7.9.tar.xz
downloading: sources/gdb-7.9.tar.xz - 17.0MB of 17.0MB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0001-sim-erc32-Disassembly-in-stand-alone-mode-did-not-wo.patch -> patches/0001-sim-erc32-Disassembly-in-stand-alone-mode-did-not-wo.patch
downloading: patches/0001-sim-erc32-Disassembly-in-stand-alone-mode-did-not-wo.patch - 1.9kB of 1.9kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0002-sim-erc32-Corrected-wrong-CPU-implementation-and-ver.patch -> patches/0002-sim-erc32-Corrected-wrong-CPU-implementation-and-ver.patch
downloading: patches/0002-sim-erc32-Corrected-wrong-CPU-implementation-and-ver.patch - 827.0 bytes of 827.0 bytes (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0003-sim-erc32-Perform-pseudo-init-if-binary-linked-to-no.patch -> patches/0003-sim-erc32-Perform-pseudo-init-if-binary-linked-to-no.patch
downloading: patches/0003-sim-erc32-Perform-pseudo-init-if-binary-linked-to-no.patch - 2.6kB of 2.6kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0004-sim-erc32-Use-fenv.h-for-host-FPU-access.patch -> patches/0004-sim-erc32-Use-fenv.h-for-host-FPU-access.patch
downloading: patches/0004-sim-erc32-Use-fenv.h-for-host-FPU-access.patch - 4.9kB of 4.9kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0005-sim-erc32-Remove-unused-defines-in-Makefile-and-swit.patch -> patches/0005-sim-erc32-Remove-unused-defines-in-Makefile-and-swit.patch
downloading: patches/0005-sim-erc32-Remove-unused-defines-in-Makefile-and-swit.patch - 871.0 bytes of 871.0 bytes (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0006-sim-erc32-Fix-incorrect-simulator-performance-report.patch -> patches/0006-sim-erc32-Fix-incorrect-simulator-performance-report.patch
downloading: patches/0006-sim-erc32-Fix-incorrect-simulator-performance-report.patch - 5.6kB of 5.6kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0007-sim-erc32-File-loading-via-command-line-did-not-work.patch -> patches/0007-sim-erc32-File-loading-via-command-line-did-not-work.patch
downloading: patches/0007-sim-erc32-File-loading-via-command-line-did-not-work.patch - 1.0kB of 1.0kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0008-sim-erc32-Added-v-command-line-switch-for-verbose-ou.patch -> patches/0008-sim-erc32-Added-v-command-line-switch-for-verbose-ou.patch
downloading: patches/0008-sim-erc32-Added-v-command-line-switch-for-verbose-ou.patch - 3.6kB of 3.6kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0009-sim-erc32-Removed-type-mismatch-compiler-warnings.patch -> patches/0009-sim-erc32-Removed-type-mismatch-compiler-warnings.patch
downloading: patches/0009-sim-erc32-Removed-type-mismatch-compiler-warnings.patch - 1.9kB of 1.9kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0010-sim-erc32-Switched-emulated-memory-to-host-endian-or.patch -> patches/0010-sim-erc32-Switched-emulated-memory-to-host-endian-or.patch
downloading: patches/0010-sim-erc32-Switched-emulated-memory-to-host-endian-or.patch - 16.0kB of 16.0kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0011-sim-erc32-use-SIM_AC_OPTION_HOSTENDIAN-to-probe-for-.patch -> patches/0011-sim-erc32-use-SIM_AC_OPTION_HOSTENDIAN-to-probe-for-.patch
downloading: patches/0011-sim-erc32-use-SIM_AC_OPTION_HOSTENDIAN-to-probe-for-.patch - 14.8kB of 14.8kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0012-sim-erc32-Use-memory_iread-function-for-instruction-.patch -> patches/0012-sim-erc32-Use-memory_iread-function-for-instruction-.patch
downloading: patches/0012-sim-erc32-Use-memory_iread-function-for-instruction-.patch - 3.8kB of 3.8kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0013-sim-erc32-Fix-a-few-compiler-warnings.patch-> patches/0013-sim-erc32-Fix-a-few-compiler-warnings.patch
downloading: patches/0013-sim-erc32-Fix-a-few-compiler-warnings.patch - 2.2kB of 2.2kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0014-sim-erc32-Use-gdb-callback-for-UART-I-O-when-linked-.patch -> patches/0014-sim-erc32-Use-gdb-callback-for-UART-I-O-when-linked-.patch
downloading: patches/0014-sim-erc32-Use-gdb-callback-for-UART-I-O-when-linked-.patch - 9.2kB of 9.2kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0015-sim-erc32-Access-memory-subsystem-through-struct-mem.patch -> patches/0015-sim-erc32-Access-memory-subsystem-through-struct-mem.patch
downloading: patches/0015-sim-erc32-Access-memory-subsystem-through-struct-mem.patch - 22.9kB of 22.9kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0016-sim-erc32-Use-readline.h-for-readline-types-and-func.patch -> patches/0016-sim-erc32-Use-readline.h-for-readline-types-and-func.patch
downloading: patches/0016-sim-erc32-Use-readline.h-for-readline-types-and-func.patch - 1.5kB of 1.5kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0017-sim-erc32-Move-local-extern-declarations-into-sis.h.patch -> patches/0017-sim-erc32-Move-local-extern-declarations-into-sis.h.patch
downloading: patches/0017-sim-erc32-Move-local-extern-declarations-into-sis.h.patch - 5.8kB of 5.8kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0018-sim-erc32-Add-support-for-LEON3-processor-emulation.patch -> patches/0018-sim-erc32-Add-support-for-LEON3-processor-emulation.patch
downloading: patches/0018-sim-erc32-Add-support-for-LEON3-processor-emulation.patch - 66.7kB of 66.7kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0019-sim-erc32-Add-support-for-LEON2-processor-emulation.patch -> patches/0019-sim-erc32-Add-support-for-LEON2-processor-emulation.patch
downloading: patches/0019-sim-erc32-Add-support-for-LEON2-processor-emulation.patch - 26.1kB of 26.1kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0020-sim-erc32-Updated-documentation.patch -> patches/0020-sim-erc32-Updated-documentation.patch
downloading: patches/0020-sim-erc32-Updated-documentation.patch - 16.1kB of 16.1kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0021-sim-erc32-Add-data-watchpoint-support.patch -> patches/0021-sim-erc32-Add-data-watchpoint-support.patch
downloading: patches/0021-sim-erc32-Add-data-watchpoint-support.patch - 10.1kB of 10.1kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0022-Add-watchpoint-support-to-gdb-simulator-interface.patch -> patches/0022-Add-watchpoint-support-to-gdb-simulator-interface.patch
downloading: patches/0022-Add-watchpoint-support-to-gdb-simulator-interface.patch - 25.5kB of 25.5kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/0023-sim-erc32-ELF-loading-could-fail-on-unaligned-sectio.patch -> patches/0023-sim-erc32-ELF-loading-could-fail-on-unaligned-sectio.patch
downloading: patches/0023-sim-erc32-ELF-loading-could-fail-on-unaligned-sectio.patch - 1.3kB of 1.3kB (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/gdb-sim-arange-inline.diff -> patches/gdb-sim-arange-inline.diff
downloading: patches/gdb-sim-arange-inline.diff - 761.0 bytes of 761.0 bytes (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/gdb-sim-cgen-inline.diff -> patches/gdb-sim-cgen-inline.diff
downloading: patches/gdb-sim-cgen-inline.diff - 706.0 bytes of 706.0 bytes (100%)
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/sources/patch-gdb-python-python-config.py -> patches/patch-gdb-python-python-config.py
downloading: patches/patch-gdb-python-python-config.py - 449.0 bytes of 449.0 bytes (100%)
building: sparc-rtems4.11-gdb-7.9-x86_64-freebsd10.1-1
reporting: tools/rtems-gdb-7.9-1.cfg ->
sparc-rtems4.11-gdb-7.9-x86_64-freebsd10.1-1.txt
reporting: tools/rtems-gdb-7.9-1.cfg ->
sparc-rtems4.11-gdb-7.9-x86_64-freebsd10.1-1.xml
config: tools/rtems-tools-4.11-1.cfg
package: rtems-tools-4.11.0-1
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/rtems-tools-4.11.0.tar.xz -> sources/rtems-tools-4.11.0.tar.xz
downloading: sources/rtems-tools-4.11.0.tar.xz - 1.6MB of 1.6MB (100%)
building: rtems-tools-4.11.0-1
reporting: tools/rtems-tools-4.11-1.cfg -> rtems-tools-4.11.0-1.txt
reporting: tools/rtems-tools-4.11-1.cfg -> rtems-tools-4.11.0-1.xml
config: tools/rtems-kernel-4.11.cfg
package: sparc-rtems4.11-kernel-4.11.0-1
download: ftp://ftp.rtems.org/pub/rtems/releases/4.11/4.11.0/rtems-4.11.0.tar.xz -> sources/rtems-4.11.0.tar.xz
downloading: sources/rtems-4.11.0.tar.xz - 9.3MB of 9.3MB (100%)
building: sparc-rtems4.11-kernel-4.11.0-1
reporting: tools/rtems-kernel-4.11.cfg -> sparc-rtems4.11-kernel-4.11.0-1.txt
reporting: tools/rtems-kernel-4.11.cfg -> sparc-rtems4.11-kernel-4.11.0-1.xml
installing: expat-2.1.0-x86_64-freebsd10.1-1 -> /opt/work/rtems/4.11.0
installing: sparc-rtems4.11-binutils-2.26-x86_64-freebsd10.1-1 -> /opt/work/rtems/4.11.0
installing: sparc-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-freebsd10.1-1 -> /opt/work/rtems/4.11.0
installing: sparc-rtems4.11-gdb-7.9-x86_64-freebsd10.1-1 -> /opt/work/rtems/4.11.0
installing: rtems-tools-4.11.0-1 -> /opt/work/rtems/4.11.0
installing: sparc-rtems4.11-kernel-4.11.0-1 -> /opt/work/rtems/4.11.0
cleaning: expat-2.1.0-x86_64-freebsd10.1-1
cleaning: sparc-rtems4.11-binutils-2.26-x86_64-freebsd10.1-1
cleaning: sparc-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-freebsd10.1-1
cleaning: sparc-rtems4.11-gdb-7.9-x86_64-freebsd10.1-1
cleaning: rtems-tools-4.11.0-1
cleaning: sparc-rtems4.11-kernel-4.11.0-1
Build Set: Time 0:19:15.713662
You can now build a third-party library or an application as defaulted in TBD.
5.2. Developer (Unstable)¶
RTEMS provides open access to it’s development processes. We call this the developer set up. The project encourages all users to inspect, review, comment and contribute to the code base. The processes described here are the same processes the core development team use when developing and maintaining RTEMS.
Warning
The development version is not for use in production and it can break from time to time.
Please read Host Computer before continuing. The following procedure assumes you have installed and configured your host operating system. It also assumes you have installed any dependent packages needed when building the tools and the kernel.
You need to select a location to build and install the RTEMS Tool chain and RTEMS. Make sure there is plenty of disk space and a fast disk is recommended. Our procedure will document building and installing the tools in a home directory called development/rtems. Using a home directory means you can do this without needing to be root. You can also use /opt/rtems/build if you have access to that path.
The location used to install the tools and kernel is called the prefix. It is best to have a prefix for each different version of RTEMS you are using. If you are using RTEMS 4.11 in production it is not a good idea to install a development version of 5 over the top. A separate prefix for each version avoids this.
The RTEMS tool chain changes less often than the RTEMS kernel. One method of working with development releases is to have a separate prefix for the RTEMS tools and a different one for the RTEMS kernel. You can then update each without interacting with the other. You can also have a number of RTEMS versions available to test with.
5.2.1. POSIX and OS X Host Tools Chain¶
This procedure will build a SPARC tool chain.
Clone the RTEMS Source Builder (RSB) repository:
$ cd
$ mkdir -p development/rtems
$ cd development/rtems
$ git clone git://git.rtems.org/rtems-source-builder.git rsb
Cloning into 'rsb'...
remote: Counting objects: 5837, done.
remote: Compressing objects: 100% (2304/2304), done.
remote: Total 5837 (delta 4014), reused 5056 (delta 3494)
Receiving objects: 100% (5837/5837), 2.48 MiB | 292.00 KiB/s, done.
Resolving deltas: 100% (4014/4014), done.
Checking connectivity... done.
Check all the host packages you need are present. Current libraries are not checked and this includes checking for the python development libraries GDB requires:
$ cd rsb
$ ./source-builder/sb-check
RTEMS Source Builder - Check, 5 (089327b5dcf9)
Environment is ok
If you are unsure how to specify the build set for the architecture you wish to build, just ask the tool:
$ ../source-builder/sb-set-builder --list-bsets <1>
RTEMS Source Builder - Set Builder, v4.11.0
Examining: config
Examining: ../source-builder/config <2>
4.10/rtems-all.bset <3>
4.10/rtems-arm.bset <4>
4.10/rtems-autotools.bset
4.10/rtems-avr.bset
4.10/rtems-bfin.bset
4.10/rtems-h8300.bset
4.10/rtems-i386.bset
4.10/rtems-lm32.bset
4.10/rtems-m32c.bset
4.10/rtems-m32r.bset
4.10/rtems-m68k.bset
4.10/rtems-mips.bset
4.10/rtems-nios2.bset
4.10/rtems-powerpc.bset
4.10/rtems-sh.bset
4.10/rtems-sparc.bset
4.11/rtems-all.bset
4.11/rtems-arm.bset
4.11/rtems-autotools.bset
4.11/rtems-avr.bset
4.11/rtems-bfin.bset
4.11/rtems-h8300.bset
4.11/rtems-i386.bset
4.11/rtems-lm32.bset
4.11/rtems-m32c.bset
4.11/rtems-m32r.bset
4.11/rtems-m68k.bset
4.11/rtems-microblaze.bset
4.11/rtems-mips.bset
4.11/rtems-moxie.bset
4.11/rtems-nios2.bset
4.11/rtems-powerpc.bset
4.11/rtems-sh.bset
4.11/rtems-sparc.bset
4.11/rtems-sparc64.bset
4.11/rtems-v850.bset
4.9/rtems-all.bset
4.9/rtems-arm.bset
4.9/rtems-autotools.bset
4.9/rtems-i386.bset
4.9/rtems-m68k.bset
4.9/rtems-mips.bset
4.9/rtems-powerpc.bset
4.9/rtems-sparc.bset
gnu-tools-4.6.bset
rtems-4.10-base.bset <5>
rtems-4.11-base.bset
rtems-4.9-base.bset
rtems-base.bset <5>
Items:
Only option required is
--list-bsetsThe paths inspected. See Configuration.
A build set to build all RTEMS 4.10 supported architectures.
The build set for the ARM architecture on RTEMS 4.10.
Support build set file with common functionality included by other build set files.
Build a tool chain for the SPARC architecture. We are using the SPARC architecture because GDB has a good simulator that lets us run and test the samples RTEMS builds by default. The current development version is 5 and is on master:
$ cd rtems
$ ../source-builder/sb-set-builder --prefix=/usr/home/chris/development/rtems/5 5/rtems-sparc
RTEMS Source Builder - Set Builder, 5 (089327b5dcf9)
Build Set: 5/rtems-sparc
Build Set: 5/rtems-autotools.bset
Build Set: 5/rtems-autotools-internal.bset
config: tools/rtems-autoconf-2.69-1.cfg
package: autoconf-2.69-x86_64-linux-gnu-1
Creating source directory: sources
download: ftp://ftp.gnu.org/gnu/autoconf/autoconf-2.69.tar.gz -> sources/autoconf-2.69.tar.gz
downloading: sources/autoconf-2.69.tar.gz - 1.8MB of 1.8MB (100%)
building: autoconf-2.69-x86_64-linux-gnu-1
config: tools/rtems-automake-1.12.6-1.cfg
package: automake-1.12.6-x86_64-linux-gnu-1
download: ftp://ftp.gnu.org/gnu/automake/automake-1.12.6.tar.gz -> sources/automake-1.12.6.tar.gz
downloading: sources/automake-1.12.6.tar.gz - 2.0MB of 2.0MB (100%)
Creating source directory: patches
download: https://git.rtems.org/rtems-tools/plain/tools/5/automake/automake-1.12.6-bugzilla.redhat.com-1239379.diff -> patches/automake-1.12.6-bugzilla.redhat.com-1239379.diff
downloading: patches/automake-1.12.6-bugzilla.redhat.com-1239379.diff - 408.0 bytes of 408.0 bytes (100%)
building: automake-1.12.6-x86_64-linux-gnu-1
cleaning: autoconf-2.69-x86_64-linux-gnu-1
cleaning: automake-1.12.6-x86_64-linux-gnu-1
Build Set: Time 0:00:12.713221
Build Set: 5/rtems-autotools-base.bset
config: tools/rtems-autoconf-2.69-1.cfg
package: autoconf-2.69-x86_64-linux-gnu-1
building: autoconf-2.69-x86_64-linux-gnu-1
reporting: tools/rtems-autoconf-2.69-1.cfg -> autoconf-2.69-x86_64-linux-gnu-1.txt
reporting: tools/rtems-autoconf-2.69-1.cfg -> autoconf-2.69-x86_64-linux-gnu-1.xml
config: tools/rtems-automake-1.12.6-1.cfg
package: automake-1.12.6-x86_64-linux-gnu-1
building: automake-1.12.6-x86_64-linux-gnu-1
reporting: tools/rtems-automake-1.12.6-1.cfg -> automake-1.12.6-x86_64-linux-gnu-1.txt
reporting: tools/rtems-automake-1.12.6-1.cfg -> automake-1.12.6-x86_64-linux-gnu-1.xml
installing: autoconf-2.69-x86_64-linux-gnu-1 -> /usr/home/chris/development/rtems/5
installing: automake-1.12.6-x86_64-linux-gnu-1 -> /usr/home/chris/development/rtems/5
cleaning: autoconf-2.69-x86_64-linux-gnu-1
cleaning: automake-1.12.6-x86_64-linux-gnu-1
Build Set: Time 0:00:09.105363
Build Set: Time 0:00:21.822083
config: devel/expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-linux-gnu-1
download: http://downloads.sourceforge.net/project/expat/expat/2.1.0/expat-2.1.0.tar.gz -> sources/expat-2.1.0.tar.gz
redirect: https://vorboss.dl.sourceforge.net/project/expat/expat/2.1.0/expat-2.1.0.tar.gz
downloading: sources/expat-2.1.0.tar.gz - 549.4kB of 549.4kB (100%)
building: expat-2.1.0-x86_64-linux-gnu-1
reporting: devel/expat-2.1.0-1.cfg -> expat-2.1.0-x86_64-linux-gnu-1.txt
reporting: devel/expat-2.1.0-1.cfg -> expat-2.1.0-x86_64-linux-gnu-1.xml
config: tools/rtems-binutils-2.29-1.cfg
package: sparc-rtems5-binutils-2.29-x86_64-linux-gnu-1
download: ftp://ftp.gnu.org/gnu/binutils/binutils-2.29.tar.bz2 -> sources/binutils-2.29.tar.bz2
downloading: sources/binutils-2.29.tar.bz2 - 27.7MB of 27.7MB (100%)
download: https://devel.rtems.org/raw-attachment/ticket/3091/0001-Fix-Binutils-2.29-PR21884.patch -> patches/0001-Fix-Binutils-2.29-PR21884.patch
downloading: patches/0001-Fix-Binutils-2.29-PR21884.patch - 8.8kB of 8.8kB (100%)
building: sparc-rtems5-binutils-2.29-x86_64-linux-gnu-1
reporting: tools/rtems-binutils-2.29-1.cfg -> sparc-rtems5-binutils-2.29-x86_64-linux-gnu-1.txt
reporting: tools/rtems-binutils-2.29-1.cfg -> sparc-rtems5-binutils-2.29-x86_64-linux-gnu-1.xml
config: tools/rtems-gcc-7.2.0-newlib-2.5.0.20170922-1.cfg
package: sparc-rtems5-gcc-7.2.0-newlib-2.5.0.20170922-x86_64-linux-gnu-1
download: https://ftp.gnu.org/gnu/gcc/gcc-7.2.0/gcc-7.2.0.tar.xz -> sources/gcc-7.2.0.tar.xz
downloading: sources/gcc-7.2.0.tar.xz - 59.4MB of 59.4MB (100%)
download: https://gcc.gnu.org/git/?p=gcc.git;a=commitdiff_plain;h=62ffbcb7502f0ff88ff7566cd6d7c59c0483ecc0 -> patches/gcc-62ffbcb7502f0ff88ff7566cd6d7c59c0483ecc0.patch
downloading: patches/gcc-62ffbcb7502f0ff88ff7566cd6d7c59c0483ecc0.patch - 1.8kB
download: https://gcc.gnu.org/git/?p=gcc.git;a=blobdiff_plain;f=gcc/config.gcc;h=593631849bb5e0df5cc4ff42c1a1cc34b7eec2f8;hp=a9196cd26d9ec24c2e3f6026f63348cae3734861;hb=e840389000b8339a63bee56d8b3...<see log> -> patches/gcc-593631849bb5e0df5cc4ff42c1a1cc34b7eec2f8.patch
downloading: patches/gcc-593631849bb5e0df5cc4ff42c1a1cc34b7eec2f8.patch - 806.0 bytes
download: https://gcc.gnu.org/git/?p=gcc.git;a=blobdiff_plain;f=gcc/config/rs6000/rtems.h;h=7ea9ebdb77b6a9b7060ad2362318e0e12b9058ae;hp=8a62fdcbaf321d616021c4c396619b7f56cf5ed2;hb=e840389000b8339a...<see log> -> patches/gcc-7ea9ebdb77b6a9b7060ad2362318e0e12b9058ae.patch
downloading: patches/gcc-7ea9ebdb77b6a9b7060ad2362318e0e12b9058ae.patch - 3.2kB
download: ftp://sourceware.org/pub/newlib/newlib-2.5.0.20170922.tar.gz -> sources/newlib-2.5.0.20170922.tar.gz
downloading: sources/newlib-2.5.0.20170922.tar.gz - 17.3MB of 17.3MB (100%)
download: https://devel.rtems.org/raw-attachment/ticket/2514/0001-RTEMS-Self-contained-POSIX-objects.patch -> patches/0001-RTEMS-Self-contained-POSIX-objects.patch
downloading: patches/0001-RTEMS-Self-contained-POSIX-objects.patch - 5.7kB of 5.7kB (100%)
download: https://sourceware.org/git/gitweb.cgi?p=newlib-cygwin.git;a=patch;h=c165a27c0147471977377acd8918ab3b446f947a -> patches/newlib-cygwin-git-c165a27c0147471977377acd8918ab3b446f947a.patch
downloading: patches/newlib-cygwin-git-c165a27c0147471977377acd8918ab3b446f947a.patch - 986.0 bytes
download: https://sourceware.org/git/gitweb.cgi?p=newlib-cygwin.git;a=patch;h=ce189d8afef720b0977b5cae7f9eabf5d49b530c -> patches/newlib-cygwin-git-ce189d8afef720b0977b5cae7f9eabf5d49b530c.patch
downloading: patches/newlib-cygwin-git-ce189d8afef720b0977b5cae7f9eabf5d49b530c.patch - 3.4kB
download: https://ftp.gnu.org/gnu/mpfr/mpfr-3.1.4.tar.bz2 -> sources/mpfr-3.1.4.tar.bz2
downloading: sources/mpfr-3.1.4.tar.bz2 - 1.2MB of 1.2MB (100%)
download: https://ftp.gnu.org/gnu/mpc/mpc-1.0.3.tar.gz -> sources/mpc-1.0.3.tar.gz
downloading: sources/mpc-1.0.3.tar.gz - 654.2kB of 654.2kB (100%)
download: https://ftp.gnu.org/gnu/gmp/gmp-6.1.0.tar.bz2 -> sources/gmp-6.1.0.tar.bz2
downloading: sources/gmp-6.1.0.tar.bz2 - 2.3MB of 2.3MB (100%)
building: sparc-rtems5-gcc-7.2.0-newlib-2.5.0.20170922-x86_64-linux-gnu-1
reporting: tools/rtems-gcc-7.2.0-newlib-2.5.0.20170922-1.cfg -> sparc-rtems5-gcc-7.2.0-newlib-2.5.0.20170922-x86_64-linux-gnu-1.txt
reporting: tools/rtems-gcc-7.2.0-newlib-2.5.0.20170922-1.cfg -> sparc-rtems5-gcc-7.2.0-newlib-2.5.0.20170922-x86_64-linux-gnu-1.xml
config: tools/rtems-gdb-8.0.1-1.cfg
package: sparc-rtems5-gdb-8.0.1-x86_64-linux-gnu-1
download: http://ftp.gnu.org/gnu/gdb/gdb-8.0.1.tar.xz -> sources/gdb-8.0.1.tar.xz
downloading: sources/gdb-8.0.1.tar.xz - 18.7MB of 18.7MB (100%)
download: https://gaisler.org/gdb/gdb-8.0.1-sis-leon2-leon3.diff -> patches/gdb-8.0.1-sis-leon2-leon3.diff
downloading: patches/gdb-8.0.1-sis-leon2-leon3.diff - 224.5kB of 224.5kB (100%)
building: sparc-rtems5-gdb-8.0.1-x86_64-linux-gnu-1
reporting: tools/rtems-gdb-8.0.1-1.cfg -> sparc-rtems5-gdb-8.0.1-x86_64-linux-gnu-1.txt
reporting: tools/rtems-gdb-8.0.1-1.cfg -> sparc-rtems5-gdb-8.0.1-x86_64-linux-gnu-1.xml
config: tools/rtems-tools-5-1.cfg
package: rtems-tools-HEAD-1
Creating source directory: sources/git
git: clone: git://git.rtems.org/rtems-tools.git -> sources/git/rtems-tools.git
git: reset: git://git.rtems.org/rtems-tools.git
git: fetch: git://git.rtems.org/rtems-tools.git -> sources/git/rtems-tools.git
git: checkout: git://git.rtems.org/rtems-tools.git => HEAD
git: pull: git://git.rtems.org/rtems-tools.git
building: rtems-tools-HEAD-1
reporting: tools/rtems-tools-5-1.cfg -> rtems-tools-HEAD-1.txt
reporting: tools/rtems-tools-5-1.cfg -> rtems-tools-HEAD-1.xml
config: tools/rtems-kernel-5.cfg
package: sparc-rtems5-kernel-5-1
building: sparc-rtems5-kernel-5-1
reporting: tools/rtems-kernel-5.cfg -> sparc-rtems5-kernel-5-1.txt
reporting: tools/rtems-kernel-5.cfg -> sparc-rtems5-kernel-5-1.xml
installing: expat-2.1.0-x86_64-linux-gnu-1 -> /usr/home/chris/development/rtems/5
installing: sparc-rtems5-binutils-2.29-x86_64-linux-gnu-1 -> /usr/home/chris/development/rtems/5
installing: sparc-rtems5-gcc-7.2.0-newlib-2.5.0.20170922-x86_64-linux-gnu-1 -> /usr/home/chris/development/rtems/5
installing: sparc-rtems5-gdb-8.0.1-x86_64-linux-gnu-1 -> /usr/home/chris/development/rtems/5
installing: rtems-tools-HEAD-1 -> /usr/home/chris/development/rtems/5
installing: sparc-rtems5-kernel-5-1 -> /usr/home/chris/development/rtems/5
cleaning: expat-2.1.0-x86_64-linux-gnu-1
cleaning: sparc-rtems5-binutils-2.29-x86_64-linux-gnu-1
cleaning: sparc-rtems5-gcc-7.2.0-newlib-2.5.0.20170922-x86_64-linux-gnu-1
cleaning: sparc-rtems5-gdb-8.0.1-x86_64-linux-gnu-1
cleaning: rtems-tools-HEAD-1
cleaning: sparc-rtems5-kernel-5-1
Build Set: Time 0:39:33.988995
5.2.2. Windows Host Tool Chain¶
This section details how you create an RTEMS development environment on Windows. The installation documented here is on Windows 7 64bit Professional. Building on Windows 10 has been reported as working.
Please see Microsoft Windows before continuing.
Note
If the RSB reports error: no hosts defaults found; please add you have probably opened an MSYS2 32bit Shell. Close all 32bit Shell windows and open the MSYS2 64bit Shell.
5.2.2.1. RTEMS Windows Tools¶
Create a workspace for RTEMS using the following shell command:
~
$ mkdir -p /c/opt/rtems
The /c path is an internal MSYS2 mount point of the C: drive. The command creates the RTEMS work space on the C: drive. If you wish to use another drive please subsitute /c with your drive letter.
We build and install all RTEMS packages under the prefix we just created. Change to that directory and get a copy of the RSB:
~
$ cd /c/opt/rtems
/c/opt/rtems
$ git clone git://git.rtems.org/rtems-source-builder.git rsb
Cloning into 'rsb'...
remote: Counting objects: 5716, done.
remote: Compressing objects: 100% (2183/2183), done.
remote: Total 5716 (delta 3919), reused 5071 (delta 3494)
Receiving objects: 100% (5716/5716), 2.46 MiB | 656.00 KiB/s, done.
Resolving deltas: 100% (3919/3919), done.
Checking connectivity... done.
Checking out files: 100% (630/630), done.
/c/opt/rtems
$ cd rsb
We are building RTEMS 4.11 tools so select the 4.11 branch:
/c/opt/rtems/rsb
$ git checkout 4.11
Branch 4.11 set up to track remote branch 4.11 from origin.
Switched to a new branch '4.11'
/c/opt/rtems/rsb
$
Check the RSB has a valid environment:
/c/opt/rtems/rsb
$ cd rtems
/c/opt/rtems/rsb/rtems
$ ../source-builder/sb-check
RTEMS Source Builder - Check, 4.11 (01ac76f2f90f)
Environment is ok
/c/opt/rtems/rsb/rtems
$
To build a set of RTEMS tools for the Intel i386 architecture. The build runs a single job rather than a job per CPU in your machine and will take a long time so please be patient. The RSB creates a log file containing all the build output and it will be changing size. The RSB command to build i386 tools is:
/c/opt/rtems/rsb/rtems
$ ../source-builder/sb-set-builder --prefix=/c/opt/rtems/4.11 \
--jobs=none 4.11/rtems-i386
RTEMS Source Builder - Set Builder, 4.11 (01ac76f2f90f)
Build Set: 4.11/rtems-i386
Build Set: 4.11/rtems-autotools.bset
Build Set: 4.11/rtems-autotools-internal.bset
config: tools/rtems-autoconf-2.69-1.cfg
package: autoconf-2.69-x86_64-w64-mingw32-1
Creating source directory: sources
download: ftp://ftp.gnu.org/gnu/autoconf/autoconf-2.69.tar.gz -> sources/autoconf-2.69.tar.gz
downloading: sources/autoconf-2.69.tar.gz - 1.8MB of 1.8MB (100%)
building: autoconf-2.69-x86_64-w64-mingw32-1
config: tools/rtems-automake-1.12.6-1.cfg
package: automake-1.12.6-x86_64-w64-mingw32-1
download: ftp://ftp.gnu.org/gnu/automake/automake-1.12.6.tar.gz -> sources/automake-1.12.6.tar.gz
downloading: sources/automake-1.12.6.tar.gz - 2.0MB of 2.0MB (100%)
building: automake-1.12.6-x86_64-w64-mingw32-1
cleaning: autoconf-2.69-x86_64-w64-mingw32-1
cleaning: automake-1.12.6-x86_64-w64-mingw32-1
Build Set: Time 0:00:42.515625
Build Set: 4.11/rtems-autotools-base.bset
config: tools/rtems-autoconf-2.69-1.cfg
package: autoconf-2.69-x86_64-w64-mingw32-1
building: autoconf-2.69-x86_64-w64-mingw32-1
reporting: tools/rtems-autoconf-2.69-1.cfg -> autoconf-2.69-x86_64-w64-mingw32-1.txt
reporting: tools/rtems-autoconf-2.69-1.cfg -> autoconf-2.69-x86_64-w64-mingw32-1.xml
config: tools/rtems-automake-1.12.6-1.cfg
package: automake-1.12.6-x86_64-w64-mingw32-1
building: automake-1.12.6-x86_64-w64-mingw32-1
reporting: tools/rtems-automake-1.12.6-1.cfg -> automake-1.12.6-x86_64-w64-mingw32-1.txt
reporting: tools/rtems-automake-1.12.6-1.cfg -> automake-1.12.6-x86_64-w64-mingw32-1.xml
tarball: tar/rtems-4.11-autotools-x86_64-w64-mingw32-1.tar.bz2
installing: autoconf-2.69-x86_64-w64-mingw32-1 -> C:\opt\rtems\4.11
installing: automake-1.12.6-x86_64-w64-mingw32-1 -> C:\opt\rtems\4.11
cleaning: autoconf-2.69-x86_64-w64-mingw32-1
cleaning: automake-1.12.6-x86_64-w64-mingw32-1
Build Set: Time 0:00:37.718750
Build Set: Time 0:01:20.234375
config: devel/expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-w64-mingw32-1
download: http://downloads.sourceforge.net/project/expat/expat/2.1.0/expat-2.1.0.tar.gz -> sources/expat-2.1.0.tar.gz
redirect: http://iweb.dl.sourceforge.net/project/expat/expat/2.1.0/expat-2.1.0.tar.gz
downloading: sources/expat-2.1.0.tar.gz - 549.4kB of 549.4kB (100%)
building: expat-2.1.0-x86_64-w64-mingw32-1
reporting: devel/expat-2.1.0-1.cfg -> expat-2.1.0-x86_64-w64-mingw32-1.txt
reporting: devel/expat-2.1.0-1.cfg -> expat-2.1.0-x86_64-w64-mingw32-1.xml
config: tools/rtems-binutils-2.24-1.cfg
package: i386-rtems4.11-binutils-2.24-x86_64-w64-mingw32-1
download: ftp://ftp.gnu.org/gnu/binutils/binutils-2.24.tar.bz2 -> sources/binutils-2.24.tar.bz2
downloading: sources/binutils-2.24.tar.bz2 - 21.7MB of 21.7MB (100%)
building: i386-rtems4.11-binutils-2.24-x86_64-w64-mingw32-1
reporting: tools/rtems-binutils-2.24-1.cfg -> i386-rtems4.11-binutils-2.24-x86_64-w64-mingw32-1.txt
reporting: tools/rtems-binutils-2.24-1.cfg -> i386-rtems4.11-binutils-2.24-x86_64-w64-mingw32-1.xml
config: tools/rtems-gcc-4.9.3-newlib-2.2.0-20150423-1.cfg
package: i386-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-w64-mingw32-1
download: ftp://ftp.gnu.org/gnu/gcc/gcc-4.9.3/gcc-4.9.3.tar.bz2 -> sources/gcc-4.9.3.tar.bz2
downloading: sources/gcc-4.9.3.tar.bz2 - 85.8MB of 85.8MB (100%)
download: ftp://sourceware.org/pub/newlib/newlib-2.2.0.20150423.tar.gz -> sources/newlib-2.2.0.20150423.tar.gz
downloading: sources/newlib-2.2.0.20150423.tar.gz - 16.7MB of 16.7MB (100%)
download: http://www.mpfr.org/mpfr-3.0.1/mpfr-3.0.1.tar.bz2 -> sources/mpfr-3.0.1.tar.bz2
downloading: sources/mpfr-3.0.1.tar.bz2 - 1.1MB of 1.1MB (100%)
download: http://www.multiprecision.org/mpc/download/mpc-0.8.2.tar.gz -> sources/mpc-0.8.2.tar.gz
downloading: sources/mpc-0.8.2.tar.gz - 535.5kB of 535.5kB (100%)
download: ftp://ftp.gnu.org/gnu/gmp/gmp-5.0.5.tar.bz2 -> sources/gmp-5.0.5.tar.bz2
downloading: sources/gmp-5.0.5.tar.bz2 - 2.0MB of 2.0MB (100%)
building: i386-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-w64-mingw32-1
reporting: tools/rtems-gcc-4.9.3-newlib-2.2.0-20150423-1.cfg ->
i386-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-w64-mingw32-1.txt
reporting: tools/rtems-gcc-4.9.3-newlib-2.2.0-20150423-1.cfg ->
i386-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-w64-mingw32-1.xml
config: tools/rtems-gdb-7.9-1.cfg
package: i386-rtems4.11-gdb-7.9-x86_64-w64-mingw32-1
download: http://ftp.gnu.org/gnu/gdb/gdb-7.9.tar.xz -> sources/gdb-7.9.tar.xz
downloading: sources/gdb-7.9.tar.xz - 17.0MB of 17.0MB (100%)
download: https://git.rtems.org/rtems-tools/plain/tools/4.11/gdb/gdb-sim-arange-inline.diff -> patches/gdb-sim-arange-inline.diff
downloading: patches/gdb-sim-arange-inline.diff - 761.0 bytes of 761.0 bytes (100%)
download: https://git.rtems.org/rtems-tools/plain/tools/4.11/gdb/gdb-sim-cgen-inline.diff -> patches/gdb-sim-cgen-inline.diff
downloading: patches/gdb-sim-cgen-inline.diff - 706.0 bytes of 706.0 bytes (100%)
building: i386-rtems4.11-gdb-7.9-x86_64-w64-mingw32-1
reporting: tools/rtems-gdb-7.9-1.cfg ->
i386-rtems4.11-gdb-7.9-x86_64-w64-mingw32-1.txt
reporting: tools/rtems-gdb-7.9-1.cfg ->
i386-rtems4.11-gdb-7.9-x86_64-w64-mingw32-1.xml
config: tools/rtems-tools-4.11-1.cfg
package: rtems-tools-4.11-1
Creating source directory: sources/git
git: clone: git://git.rtems.org/rtems-tools.git -> sources/git/rtems-tools.git
git: reset: git://git.rtems.org/rtems-tools.git
git: fetch: git://git.rtems.org/rtems-tools.git -> sources/git/rtems-tools.git
git: checkout: git://git.rtems.org/rtems-tools.git => 4.11
git: pull: git://git.rtems.org/rtems-tools.git
building: rtems-tools-4.11-1
reporting: tools/rtems-tools-4.11-1.cfg -> rtems-tools-4.11-1.txt
reporting: tools/rtems-tools-4.11-1.cfg -> rtems-tools-4.11-1.xml
config: tools/rtems-kernel-4.11.cfg
installing: expat-2.1.0-x86_64-w64-mingw32-1 -> C:\opt\rtems\4.11
installing: i386-rtems4.11-binutils-2.24-x86_64-w64-mingw32-1 -> C:\opt\rtems\4.11
installing: i386-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-w64-mingw32-1 -> C:\opt\rtems\4.11
installing: i386-rtems4.11-gdb-7.9-x86_64-w64-mingw32-1 -> C:\opt\rtems\4.11
installing: rtems-tools-4.11-1 -> C:\opt\rtems\4.11
cleaning: expat-2.1.0-x86_64-w64-mingw32-1
cleaning: i386-rtems4.11-binutils-2.24-x86_64-w64-mingw32-1
cleaning: i386-rtems4.11-gcc-4.9.3-newlib-2.2.0.20150423-x86_64-w64-mingw32-1
cleaning: i386-rtems4.11-gdb-7.9-x86_64-w64-mingw32-1
cleaning: rtems-tools-4.11-1
Build Set: Time 1:32:58.972919
/c/opt/rtems/rsb/rtems
$
5.2.2.2. Building the Kernel¶
We can now build the RTEMS kernel using the RTEMS tools we have just built. First we need to set the path to the tools:
/c
$ cd /c/opt/rtems
/c/opt/rtems
$ export PATH=/c/opt/rtems/4.11/bin:$PATH
/c/opt/rtems
$
We currently build RTEMS from the git release branch for 4.11:
/c/opt/rtems
$ mkdir kernel
/c/opt/rtems
$ cd kernel
/c/opt/rtems/kernel
$ git clone git://git.rtems.org/rtems.git rtems
Cloning into 'rtems'...
remote: Counting objects: 482766, done.
remote: Compressing objects: 100% (88781/88781), done.
remote: Total 482766 (delta 389610), reused 475155 (delta 383437)
Receiving objects: 100% (482766/482766), 69.77 MiB | 697.00 KiB/s, done.
Resolving deltas: 100% (389610/389610), done.
Checking connectivity... done.
Checking out files: 100% (10626/10626), done.
/c/opt/rtems/kernel
$ cd rtems
/c/opt/rtems/kernel/rtems
$ git checkout 4.11
Checking out files: 100% (2553/2553), done.
Branch 4.11 set up to track remote branch 4.11 from origin.
Switched to a new branch '4.11'
/c/opt/rtems/kernel
$
The kernel code cloned from git needs to be bootstrapped. Bootstrapping creates autoconf and automake generated files. To bootstrap we first clean away any files, then generate the pre-install header file lists and finally we generate the autoconf and automake files using the RSB’s bootstrap tool. First we clean any generated files that exist:
/c/opt/rtems/kernel/rtems
$ ./bootstrap -c
removing automake generated Makefile.in files
removing configure files
removing aclocal.m4 files
Then we generate the pre-install header file automake make files:
/c/opt/rtems/kernel/rtems
$ ./bootstrap -p
Generating ./c/src/ada/preinstall.am
Generating ./c/src/lib/libbsp/arm/altera-cyclone-v/preinstall.am
Generating ./c/src/lib/libbsp/arm/atsam/preinstall.am
Generating ./c/src/lib/libbsp/arm/beagle/preinstall.am
Generating ./c/src/lib/libbsp/arm/csb336/preinstall.am
Generating ./c/src/lib/libbsp/arm/csb337/preinstall.am
Generating ./c/src/lib/libbsp/arm/edb7312/preinstall.am
.......
Generating ./cpukit/score/cpu/mips/preinstall.am
Generating ./cpukit/score/cpu/moxie/preinstall.am
Generating ./cpukit/score/cpu/nios2/preinstall.am
Generating ./cpukit/score/cpu/no_cpu/preinstall.am
Generating ./cpukit/score/cpu/or1k/preinstall.am
Generating ./cpukit/score/cpu/powerpc/preinstall.am
Generating ./cpukit/score/cpu/sh/preinstall.am
Generating ./cpukit/score/cpu/sparc/preinstall.am
Generating ./cpukit/score/cpu/sparc64/preinstall.am
Generating ./cpukit/score/cpu/v850/preinstall.am
Generating ./cpukit/score/preinstall.am
Generating ./cpukit/telnetd/preinstall.am
Generating ./cpukit/wrapup/preinstall.am
Generating ./cpukit/zlib/preinstall.am
/c/opt/rtems/kernel/rtems
Finally we run the RSB’s parallel bootstrap command:
$ /c/opt/rtems/rsb/source-builder/sb-bootstrap
RTEMS Source Builder - RTEMS Bootstrap, 4.11 (76188ee494dd)
1/139: autoreconf: configure.ac
2/139: autoreconf: c/configure.ac
3/139: autoreconf: c/src/configure.ac
4/139: autoreconf: c/src/ada-tests/configure.ac
5/139: autoreconf: c/src/lib/libbsp/arm/configure.ac
6/139: autoreconf: c/src/lib/libbsp/arm/altera-cyclone-v/configure.ac
7/139: autoreconf: c/src/lib/libbsp/arm/atsam/configure.ac
8/139: autoreconf: c/src/lib/libbsp/arm/beagle/configure.ac
9/139: autoreconf: c/src/lib/libbsp/arm/csb336/configure.ac
10/139: autoreconf: c/src/lib/libbsp/arm/csb337/configure.ac
11/139: autoreconf: c/src/lib/libbsp/arm/edb7312/configure.ac
.......
129/139: autoreconf: testsuites/samples/configure.ac
130/139: autoreconf: testsuites/smptests/configure.ac
131/139: autoreconf: testsuites/sptests/configure.ac
132/139: autoreconf: testsuites/tmtests/configure.ac
133/139: autoreconf: testsuites/tools/configure.ac
134/139: autoreconf: testsuites/tools/generic/configure.ac
135/139: autoreconf: tools/build/configure.ac
136/139: autoreconf: tools/cpu/configure.ac
137/139: autoreconf: tools/cpu/generic/configure.ac
138/139: autoreconf: tools/cpu/nios2/configure.ac
139/139: autoreconf: tools/cpu/sh/configure.ac
Bootstrap time: 0:20:38.759766
/c/opt/rtems/kernel/rtems
$
We will build the RTEMS kernel for the i386 target and the pc686 BSP. You can check the available BSPs by running the rtems-bsps command found in the top directory of the RTEMS kernel source. We build the Board Support Package (BSP) outside the kernel source tree:
/c/opt/rtems/kernel/rtems
$ cd ..
/c/opt/rtems/kernel
$ mkdir pc686
/c/opt/rtems/kernel
$ cd pc686
/c/opt/rtems/kernel/pc686
$
Configure the RTEMS kernel to build pc686 BSP for the i386 target with networking disabled, We will build the external libBSD stack later:
/c/opt/rtems/kernel/pc686
$ /c/opt/rtems/kernel/rtems/configure --prefix=/c/opt/rtems/4.11 \
--target=i386-rtems4.11 --disable-networking --enable-rtemsbsp=pc686
checking for gmake... no
checking for make... make
checking for RTEMS Version... 4.11.99.0
checking build system type... x86_64-pc-mingw64
checking host system type... x86_64-pc-mingw64
checking target system type... i386-pc-rtems4.11
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... /usr/bin/mkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking that generated files are newer than configure... done
configure: creating ./config.status
configure: configuring in ./tools/build
.......
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking that generated files are newer than configure... done
configure: creating ./config.status
config.status: creating Makefile
target architecture: i386.
available BSPs: pc686.
'make all' will build the following BSPs: pc686.
other BSPs can be built with 'make RTEMS_BSP="bsp1 bsp2 ..."'
config.status: creating Makefile
/c/opt/rtems/kernel/pc686
$
Build the kernel:
/c/opt/rtems/kernel/pc686
$ make
Making all in tools/build
make[1]: Entering directory '/c/opt/rtems/kernel/pc686/tools/build'
make all-am
make[2]: Entering directory '/c/opt/rtems/kernel/pc686/tools/build'
gcc -DHAVE_CONFIG_H -I. -I/c/opt/rtems/kernel/rtems/tools/build -g -O2 -MT
cklength.o -MD -MP -MF .deps/cklength.Tpo -c -o cklength.o
/c/opt/rtems/kernel/rtems/tools/build/cklength.c
gcc -DHAVE_CONFIG_H -I. -I/c/opt/rtems/kernel/rtems/tools/build -g -O2 -MT
eolstrip.o -MD -MP -MF .deps/eolstrip.Tpo -c -o eolstrip.o
/c/opt/rtems/kernel/rtems/tools/build/eolstrip.c
..........
i386-rtems4.11-objcopy -O binary nsecs.nxe nsecs.bin
../../../../../pc686/build-tools/bin2boot -v nsecs.ralf 0x00097E00
../../../../../pc686/lib/start16.bin 0x00097C00 0 nsecs.bin 0x00100000 0
header address 0x00097e00, its memory size 0xzx
first image address 0x00097c00, its memory size 0x00000200
second image address 0x00100000, its memory size 0x0003d800
rm -f nsecs.nxe
make[6]: Leaving directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c/pc686/testsuites/samples/nsecs'
make[5]: Leaving directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c/pc686/testsuites/samples'
make[4]: Leaving directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c/pc686/testsuites/samples'
make[4]: Entering directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c/pc686/testsuites'
make[4]: Nothing to be done for 'all-am'.
make[4]: Leaving directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c/pc686/testsuites'
make[3]: Leaving directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c/pc686/testsuites'
make[2]: Leaving directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c/pc686'
make[1]: Leaving directory '/c/opt/rtems/kernel/pc686/i386-rtems4.11/c'
make[1]: Entering directory '/c/opt/rtems/kernel/pc686'
make[1]: Nothing to be done for 'all-am'.
make[1]: Leaving directory '/c/opt/rtems/kernel/pc686'
/c/opt/rtems/kernel/pc696
$
Install the kernel to our prefix:
$ make install
Making install in tools/build
make[1]: Entering directory '/c/opt/rtems/kernel/pc686/tools/build'
make[2]: Entering directory '/c/opt/rtems/kernel/pc686/tools/build'
/usr/bin/mkdir -p '/c/opt/rtems/4.11/bin'
/usr/bin/install -c cklength.exe eolstrip.exe packhex.exe unhex.exe
rtems-bin2c.exe '/c/opt/rtems/4.11/bin'
/usr/bin/mkdir -p '/c/opt/rtems/4.11/bin'
/usr/bin/install -c install-if-change '/c/opt/rtems/4.11/bin'
make[2]: Nothing to be done for 'install-data-am'.
make[2]: Leaving directory '/c/opt/rtems/kernel/pc686/tools/build'
make[1]: Leaving directory '/c/opt/rtems/kernel/pc686/tools/build'
Making install in tools/cpu
make[1]: Entering directory '/c/opt/rtems/kernel/pc686/tools/cpu'
Making install in generic
make[2]: Entering directory '/c/opt/rtems/kernel/pc686/tools/cpu/generic'
make[3]: Entering directory '/c/opt/rtems/kernel/pc686/tools/cpu/generic'
make[3]: Nothing to be done for 'install-exec-am'.
make[3]: Nothing to be done for 'install-data-am'.
make[3]: Leaving directory '/c/opt/rtems/kernel/pc686/tools/cpu/generic'
make[2]: Leaving directory '/c/opt/rtems/kernel/pc686/tools/cpu/generic'
make[2]: Entering directory '/c/opt/rtems/kernel/pc686/tools/cpu'
make[3]: Entering directory '/c/opt/rtems/kernel/pc686/tools/cpu'
make[3]: Nothing to be done for 'install-exec-am'.
make[3]: Nothing to be done for 'install-data-am'.
..........
make[2]: Entering directory '/c/opt/rtems/kernel/pc686'
make[2]: Nothing to be done for 'install-exec-am'.
/usr/bin/mkdir -p '/c/opt/rtems/4.11/make'
/usr/bin/install -c -m 644 /c/opt/rtems/kernel/rtems/make/main.cfg
/c/opt/rtems/kernel/rtems/make/leaf.cfg '/c/opt/rtems/4.11/make'
/usr/bin/mkdir -p '/c/opt/rtems/4.11/share/rtems4.11/make/Templates'
/usr/bin/install -c -m 644
/c/opt/rtems/kernel/rtems/make/Templates/Makefile.dir
/c/opt/rtems/kernel/rtems/make/Templates/Makefile.leaf
/c/opt/rtems/kernel/rtems/make/Templates/Makefile.lib
'/c/opt/rtems/4.11/share/rtems4.11/make/Templates'
/usr/bin/mkdir -p '/c/opt/rtems/4.11/make/custom'
/usr/bin/install -c -m 644 /c/opt/rtems/kernel/rtems/make/custom/default.cfg
'/c/opt/rtems/4.11/make/custom'
make[2]: Leaving directory '/c/opt/rtems/kernel/pc686'
make[1]: Leaving directory '/c/opt/rtems/kernel/pc686'
/c/opt/rtems/kernel/pc686
$
5.3. RTEMS Kernel¶
RTEMS is an open source real-time operating system. As a user you have access to all the source code. The RTEMS Kernel section will show you how you build the RTEMS kernel on your host.
5.3.1. Development Sources¶
Create a new location to build the RTEMS kernel:
$ cd
$ cd development/rtems
$ mkdir kernel
$ cd kernel
Clone the RTEMS respository:
$ git clone git://git.rtems.org/rtems.git rtems
Cloning into 'rtems'...
remote: Counting objects: 483342, done.
remote: Compressing objects: 100% (88974/88974), done.
remote: Total 483342 (delta 390053), reused 475669 (delta 383809)
Receiving objects: 100% (483342/483342), 69.88 MiB | 1.37 MiB/s, done.
Resolving deltas: 100% (390053/390053), done.
Checking connectivity... done.
5.3.2. Tools Path Set Up¶
We need to set our path to include the RTEMS tools we built in the previous section. The RTEMS tools needs to be first in your path because RTEMS provides specific versions of the autoconf and automake tools. We want to use the RTEMS version and not your host’s versions:
$ export PATH=$HOME/development/rtems/5/bin:$PATH
5.3.3. Bootstrapping¶
The developers version of the code from git requires we bootstrap the source code. This is an autoconf and automake bootstrap to create the various files generated by autoconf and automake. RTEMS does not keep these generated files under version control. The bootstrap process is slow so to speed it up the RSB provides a command that can perform the bootstrap in parallel using your available cores. We need to enter the cloned source directory then run the bootstrap commands:
$ cd rtems
$ ./bootstrap -c && $HOME/development/rtems/rsb/source-builder/sb-bootstrap
removing automake generated Makefile.in files
removing configure files
removing aclocal.m4 files
RTEMS Source Builder - RTEMS Bootstrap, 5 (089327b5dcf9)
1/139: autoreconf: configure.ac
2/139: autoreconf: cpukit/configure.ac
3/139: autoreconf: tools/cpu/configure.ac
4/139: autoreconf: tools/cpu/generic/configure.ac
5/139: autoreconf: tools/cpu/sh/configure.ac
6/139: autoreconf: tools/cpu/nios2/configure.ac
7/139: autoreconf: tools/build/configure.ac
8/139: autoreconf: doc/configure.ac
......
124/139: autoreconf: c/src/make/configure.ac
125/139: autoreconf: c/src/librtems++/configure.ac
126/139: autoreconf: c/src/ada-tests/configure.ac
127/139: autoreconf: testsuites/configure.ac
128/139: autoreconf: testsuites/libtests/configure.ac
129/139: autoreconf: testsuites/mptests/configure.ac
130/139: autoreconf: testsuites/fstests/configure.ac
131/139: autoreconf: testsuites/sptests/configure.ac
132/139: autoreconf: testsuites/tmtests/configure.ac
133/139: autoreconf: testsuites/smptests/configure.ac
134/139: autoreconf: testsuites/tools/configure.ac
135/139: autoreconf: testsuites/tools/generic/configure.ac
136/139: autoreconf: testsuites/psxtests/configure.ac
137/139: autoreconf: testsuites/psxtmtests/configure.ac
138/139: autoreconf: testsuites/rhealstone/configure.ac
139/139: autoreconf: testsuites/samples/configure.ac
Bootstrap time: 0:02:47.398824
5.3.4. Building a BSP¶
We build RTEMS in a directory outside of the source tree we have just cloned and bootstrapped. You cannot build RTEMS while in the source tree. Lets create a suitable directory using the name of the BSP we are going to build:
$ cd ..
$ mkdir erc32
$ cd erc32
Configure RTEMS using the configure command. We use a full path to configure so the object files built contain the absolute path of the source files. If you are source level debugging you will be able to access the source code to RTEMS from the debugger. We will build for the erc32 BSP with POSIX enabled and the networking stack disabled:
$ $HOME/development/rtems/kernel/rtems/configure --prefix=$HOME/development/rtems/5 \
--target=sparc-rtems5 --enable-rtemsbsp=erc32 --enable-posix \
--disable-networking
checking for gmake... no
checking for make... make
checking for RTEMS Version... 4.11.99.0
checking build system type... x86_64-pc-linux-gnu
checking host system type... x86_64-pc-linux-gnu
checking target system type... sparc-unknown-rtems5
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... /bin/mkdir -p
checking for gawk... no
checking for mawk... mawk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking that generated files are newer than configure... done
......
checking target system type... sparc-unknown-rtems5
checking rtems target cpu... sparc
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for sparc-rtems5-strip... sparc-rtems5-strip
checking for a thread-safe mkdir -p... /bin/mkdir -p
checking for gawk... no
checking for mawk... mawk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking that generated files are newer than configure... done
configure: creating ./config.status
config.status: creating Makefile
target architecture: sparc.
available BSPs: erc32.
'make all' will build the following BSPs: erc32.
other BSPs can be built with 'make RTEMS_BSP="bsp1 bsp2 ..."'
config.status: creating Makefile
Build RTEMS using two cores:
$ make -j 2
Making all in tools/build
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/build'
make all-am
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/build'
gcc -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/tools/build -g -O2 -MT cklength.o -MD -MP -MF .deps/cklength.Tpo -c -o cklength.o /home/chris/development/rtems/kernel/rtems/tools/build/cklength.c
gcc -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/tools/build -g -O2 -MT eolstrip.o -MD -MP -MF .deps/eolstrip.Tpo -c -o eolstrip.o /home/chris/development/rtems/kernel/rtems/tools/build/eolstrip.c
mv -f .deps/cklength.Tpo .deps/cklength.Po
mv -f .deps/eolstrip.Tpo .deps/eolstrip.Po
gcc -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/tools/build -g -O2 -MT compat.o -MD -MP -MF .deps/compat.Tpo -c -o compat.o /home/chris/development/rtems/kernel/rtems/tools/build/compat.c
gcc -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/tools/build -g -O2 -MT packhex.o -MD -MP -MF .deps/packhex.Tpo -c -o packhex.o /home/chris/development/rtems/kernel/rtems/tools/build/packhex.c
mv -f .deps/compat.Tpo .deps/compat.Po
gcc -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/tools/build -g -O2 -MT unhex.o -MD -MP -MF .deps/unhex.Tpo -c -o unhex.o /home/chris/development/rtems/kernel/rtems/tools/build/unhex.c
mv -f .deps/packhex.Tpo .deps/packhex.Po
gcc -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/tools/build -g -O2 -MT rtems-bin2c.o -MD -MP -MF .deps/rtems-bin2c.Tpo -c -o rtems-bin2c.o /home/chris/development/rtems/kernel/rtems/tools/build/rtems-bin2c.c
mv -f .deps/unhex.Tpo .deps/unhex.Po
gcc -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/tools/build -g -O2 -MT binpatch.o -MD -MP -MF .deps/binpatch.Tpo -c -o binpatch.o /home/chris/development/rtems/kernel/rtems/tools/build/binpatch.c
mv -f .deps/rtems-bin2c.Tpo .deps/rtems-bin2c.Po
gcc -g -O2 -o cklength cklength.o
mv -f .deps/binpatch.Tpo .deps/binpatch.Po
gcc -g -O2 -o eolstrip eolstrip.o compat.o
gcc -g -O2 -o packhex packhex.o
gcc -g -O2 -o rtems-bin2c rtems-bin2c.o compat.o
gcc -g -O2 -o unhex unhex.o compat.o
gcc -g -O2 -o binpatch binpatch.o
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/build'
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/build'
Making all in tools/cpu
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
Making all in generic
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu/generic'
make[2]: Nothing to be done for 'all'.
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu/generic'
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
make[2]: Nothing to be done for 'all-am'.
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
Making all in testsuites/tools
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32/testsuites/tools'
Making all in generic
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/testsuites/tools/generic'
make[2]: Nothing to be done for 'all'.
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/testsuites/tools/generic'
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/testsuites/tools'
make[2]: Nothing to be done for 'all-am'.
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/testsuites/tools'
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32/testsuites/tools'
Making all in sparc-rtems5/c
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/c'
Making all in .
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/c'
Configuring RTEMS_BSP=erc32
checking for gmake... no
checking for make... make
checking build system type... x86_64-pc-linux-gnu
checking host system type... sparc-unknown-rtems5
......
sparc-rtems5-gcc -B../../../../../erc32/lib/ -specs bsp_specs -qrtems -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/c/src/../../testsuites/samples/nsecs -I.. -I/home/chris/development/rtems/kernel/rtems/c/src/../../testsuites/samples/../support/include -mcpu=cypress -O2 -g -ffunction-sections -fdata-sections -Wall -Wmissing-prototypes -Wimplicit-function-declaration -Wstrict-prototypes -Wnested-externs -MT init.o -MD -MP -MF .deps/init.Tpo -c -o init.o /home/chris/development/rtems/kernel/rtems/c/src/../../testsuites/samples/nsecs/init.c
sparc-rtems5-gcc -B../../../../../erc32/lib/ -specs bsp_specs -qrtems -DHAVE_CONFIG_H -I. -I/home/chris/development/rtems/kernel/rtems/c/src/../../testsuites/samples/nsecs -I.. -I/home/chris/development/rtems/kernel/rtems/c/src/../../testsuites/samples/../support/include -mcpu=cypress -O2 -g -ffunction-sections -fdata-sections -Wall -Wmissing-prototypes -Wimplicit-function-declaration -Wstrict-prototypes -Wnested-externs -MT empty.o -MD -MP -MF .deps/empty.Tpo -c -o empty.o /home/chris/development/rtems/kernel/rtems/c/src/../../testsuites/samples/nsecs/empty.c
mv -f .deps/empty.Tpo .deps/empty.Po
mv -f .deps/init.Tpo .deps/init.Po
sparc-rtems5-gcc -B../../../../../erc32/lib/ -specs bsp_specs -qrtems -mcpu=cypress -O2 -g -ffunction-sections -fdata-sections -Wall -Wmissing-prototypes -Wimplicit-function-declaration -Wstrict-prototypes -Wnested-externs -Wl,--gc-sections -mcpu=cypress -o nsecs.exe init.o empty.o
sparc-rtems5-nm -g -n nsecs.exe > nsecs.num
sparc-rtems5-size nsecs.exe
text data bss dec hex filename
121392 1888 6624 129904 1fb70 nsecs.exe
cp nsecs.exe nsecs.ralf
make[6]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/ c/erc32/testsuites/samples/nsecs'
make[5]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/ c/erc32/testsuites/samples'
make[4]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/ c/erc32/testsuites/samples'
make[4]: Entering directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/ c/erc32/testsuites'
make[4]: Nothing to be done for 'all-am'.
make[4]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/ c/erc32/testsuites'
make[3]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/ c/erc32/testsuites'
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/ c/erc32'
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/c'
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32'
make[1]: Nothing to be done for 'all-am'.
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32'
5.3.5. Installing A BSP¶
All that remains to be done is to install the kernel. Installing RTEMS copies the API headers and architecture specific libraries to a locaiton under the prefix you provide. You can install any number of BSPs under the same prefix. We recommend you have a separate prefix for different versions of RTEMS. Do not mix versions of RTEMS under the same prefix. Make installs RTEMS with the following command:
$ make install
Making install in tools/build
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/build'
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/build'
/bin/mkdir -p '/home/chris/development/rtems/5/bin'
/usr/bin/install -c cklength eolstrip packhex unhex rtems-bin2c '/home/chris/development/rtems/5/bin'
/bin/mkdir -p '/home/chris/development/rtems/5/bin'
/usr/bin/install -c install-if-change '/home/chris/development/rtems/5/bin'
make[2]: Nothing to be done for 'install-data-am'.
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/build'
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/build'
Making install in tools/cpu
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
Making install in generic
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu/generic'
make[3]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu/generic'
make[3]: Nothing to be done for 'install-exec-am'.
make[3]: Nothing to be done for 'install-data-am'.
make[3]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu/generic'
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu/generic'
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
make[3]: Entering directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
make[3]: Nothing to be done for 'install-exec-am'.
make[3]: Nothing to be done for 'install-data-am'.
make[3]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32/tools/cpu'
....
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32/sparc-rtems5/c'
make[1]: Entering directory '/home/chris/development/rtems/kernel/erc32'
make[2]: Entering directory '/home/chris/development/rtems/kernel/erc32'
make[2]: Nothing to be done for 'install-exec-am'.
/bin/mkdir -p '/home/chris/development/rtems/5/make'
/usr/bin/install -c -m 644 /home/chris/development/rtems/kernel/rtems/make/main.cfg /home/chris/development/rtems/kernel/rtems/make/leaf.cfg '/home/chris/development/rtems/5/make'
/bin/mkdir -p '/home/chris/development/rtems/5/share/rtems5/make/Templates'
/usr/bin/install -c -m 644 /home/chris/development/rtems/kernel/rtems/make/Templates/Makefile.dir /home/chris/development/rtems/kernel/rtems/make/Templates/Makefile.leaf /home/chris/development/rtems/kernel/rtems/make/Templates/Makefile.lib '/home/chris/development/rtems/5/share/rtems5/make/Templates'
/bin/mkdir -p '/home/chris/development/rtems/5/make/custom'
/usr/bin/install -c -m 644 /home/chris/development/rtems/kernel/rtems/make/custom/default.cfg '/home/chris/development/rtems/5/make/custom'
make[2]: Leaving directory '/home/chris/development/rtems/kernel/erc32'
make[1]: Leaving directory '/home/chris/development/rtems/kernel/erc32'
5.3.6. Contributing Patches¶
RTEMS welcomes fixes to bugs and new features. The RTEMS Project likes to have bugs fixed against a ticket created on our Developer Site. Please raise a ticket if you have a bug. Any changes that are made can be tracked against the ticket. If you want to add a new a feature please post a message to Developers Mailing List describing what you would like to implement. The RTEMS maintainer will help decide if the feature is in the best interest of the project. Not everything is and the maintainers need to evalulate how much effort it is to maintain the feature. Once accepted into the source tree it becomes the responsibility of the maintainers to keep the feature updated and working.
Changes to the source tree are tracked using git. If you have not made changes and enter the source tree and enter a git status command you will see nothing has changed:
$ cd ../rtems
$ git status
On branch master
Your branch is up-to-date with 'origin/master'.
nothing to commit, working directory clean
We will make a change to the source code. In this example I change the help message to the RTEMS shell’s halt command. Running the same git status command reports:
$ git status
On branch master
Your branch is up-to-date with 'origin/master'.
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git checkout -- <file>..." to discard changes in working directory)
modified: cpukit/libmisc/shell/main_halt.c
no changes added to commit (use "git add" and/or "git commit -a")
As an example I have a ticket open and the ticket number is 9876. I commit the change with the follow git command:
$ git commit cpukit/libmisc/shell/main_halt.c
An editor is opened and I enter my commit message. The first line is a title and the following lines form a body. My message is:
shell: Add more help detail to the halt command.
Closes #9876.
# Please enter the commit message for your changes. Lines starting
# with '#' will be ignored, and an empty message aborts the commit.
# Explicit paths specified without -i or -o; assuming --only paths...
#
# Committer: Chris Johns <chrisj@rtems.org>
#
# On branch master
# Your branch is up-to-date with 'origin/master'.
#
# Changes to be committed:
# modified: cpukit/libmisc/shell/main_halt.c
When you save and exit the editor git will report the commit’s status:
$ git commit cpukit/libmisc/shell/main_halt.c
[master 9f44dc9] shell: Add more help detail to the halt command.
1 file changed, 1 insertion(+), 1 deletion(-)
You can either email the patch to Developers Mailing List with the following git command, and it is minus one on the command line:
$ git send-email --to=devel@rtems.org -1
<add output here>
Or you can ask git to create a patch file using:
$ git format-patch -1
0001-shell-Add-more-help-detail-to-the-halt-command.patch
This patch can be attached to a ticket.
5.4. Project Sandboxing¶
Project specific sandboxes let you have a number of projects running in parallel with each project in its own sandbox. You simply have a prefix per project and under that prefix you create a simple yet repeatable structure.
As an example lets say I have a large disk mounted under /bd for Big Disk. As root create a directory called projects and give the directory suitable permissions to be writable by you as a user.
Lets create a project sandbox for my Box Sorter project. First create a project directory called /bd/projects/box-sorter. Under this create rtems and under that create rtems-4.11.0. Under this path you can follow the Releases procedure to build a tool set using the prefix of /bd/projects/box-sorter/rtems/4.11.0. You are free to create your project specific directories under /bd/projects/box-sorter. The top level directories would be:
/bd/projectsProject specific development trees.
/bd/projects/box-sorterBox Sorter project sandbox.
/bd/projects/box-sorter/rtems/4.11.0Project prefix for RTEMS 4.11.0 compiler, debuggers, tools and installed Board Support Package (BSP).
A variation is to use the --without-rtems option with the RSB to not build the BSPs when building the tools and to build RTEMS specifically for each project. This lets you have a production tools installed at a top level on your disk and each project can have a specific and possibly customised version of RTEMS. The top level directories would be:
/bd/rtemsThe top path to production tools.
/bd/rtems/4.11.0Production prefix for RTEMS 4.11.0 compiler, debuggers and tools.
/bd/projectsProject specific development trees.
/bd/projects/box-sorterBox Sorter project sandbox.
/bd/projects/box-sorter/rtemsBox Sorter project’s custom RTEMS kernel source and installed BSP.
A further varation if there is an RTEMS kernel you want to share between projects is it to move this to a top level and share. In this case you will end up with:
/bd/rtemsThe top path to production tools and kernels.
/bd/rtems/4.11.0Production prefix for RTEMS 4.11.0.
/bd/rtems/4.11.0/toolsProduction prefix for RTEMS 4.11.0 compiler, debuggers and tools.
/bd/rtems/4.11.0/bspsProduction prefix for RTEMS 4.11.0 Board Support Packages (BSPs).
/bd/projectsProject specific development trees.
/bd/projects/box-sorterBox Sorter project sandbox.
Finally you can have a single set of production tools and RTEMS BSPs on the disk under /bd/rtems you can share between your projects. The top level directories would be:
/bd/rtemsThe top path to production tools and kernels.
/bd/rtems/4.11.0Production prefix for RTEMS 4.11.0 compiler, debuggers, tools and Board Support Packages (BSPs).
/bd/projectsProject specific development trees.
/bd/projects/box-sorterBox Sorter project sandbox.
The project sandoxing approach allows you move a specific production part into the project’s sandbox to allow you to customise it. This is useful if you are testing new releases. The typical dependency is the order listed above. You can test new RTEMS kernels with production tools but new tools will require you build the kernel with them. Release notes with each release will let know what you need to update.
If the machine is a central project development machine simply replace projects with users and give each user a personal directory.
6. Target Hardware¶
6.1. Targets¶
Target hardware that can run RTEMS is often referred to simply as the target because RTEMS is specifically aimed at that target hardware. An RTEMS executable is statically linked and executes in a single address space on the target hardware. A statically linked executable means the RTEMS Kernel, drivers, third-party packages and application code is linked into a single executable image. A single address space means no virtual memory and no memory protected process address space is running within the RTEMS arena and the RTEMS executive, drivers and application have unprotected access to the whole address space and all hardware.
Target hardware supported by RTEMS has a Board Support Package or BSP. A BSP is a specific instance of an RTEMS architecture that allows the creation of an RTEMS executable. You can view the layering as:
RTEMS targets are grouped by architectures and within an architecture there are a number of Board Support Packages or BPSs. An architecture is a specific class or family of processors and can be large such as ARM or specific such as the NIOS-II or Microblaze.
RTEMS is designed to be ported to new target hardware easily and efficiently.
6.2. Architectures¶
An RTEMS architecture is a class or family of a processor architecture that RTEMS supports. The RTEMS architecture model follows the architecture model of GCC. An architecture in GCC results in a specific RTEMS GCC compiler. This compiler may support a range of processors in the family that may have differences in instructions sets, floating point support or other aspects. RTEMS configures GCC to create separate runtime libraries for each supported instruction set, floating point unit, vector unit, word size (e.g. 32-bit and 64-bit), endianess, code model, ABI, processor errata workarounds, and so on in the architecture. This is termed multilib. Multilibs are chosen automatically by GCC via selecting a specific set of machine options.
You can query the multilibs of a specific RTEMS GCC compiler via the -print-multi-lib option:
$ sparc-rtems5-gcc -print-multi-lib
.;
soft;@msoft-float
v8;@mcpu=v8
leon3;@mcpu=leon3
leon3v7;@mcpu=leon3v7
leon;@mcpu=leon
leon3/gr712rc;@mcpu=leon3@mfix-gr712rc
leon3v7/gr712rc;@mcpu=leon3v7@mfix-gr712rc
leon/ut699;@mcpu=leon@mfix-ut699
leon/at697f;@mcpu=leon@mfix-at697f
soft/v8;@msoft-float@mcpu=v8
soft/leon3;@msoft-float@mcpu=leon3
soft/leon3v7;@msoft-float@mcpu=leon3v7
soft/leon;@msoft-float@mcpu=leon
soft/leon3/gr712rc;@msoft-float@mcpu=leon3@mfix-gr712rc
soft/leon3v7/gr712rc;@msoft-float@mcpu=leon3v7@mfix-gr712rc
soft/leon/ut699;@msoft-float@mcpu=leon@mfix-ut699
soft/leon/at697f;@msoft-float@mcpu=leon@mfix-at697f
Each printed line represents a multilib. The . corresponds to the default multilib. It is used if a set of machine options does not match to a specialized multilib. The string before the ; describes the directory in the GCC installation used for the particular multilib. After the ; the set of machine options for this multilib follows separated by @ characters.
You can figure out the multilib selected by GCC for a set of machine options with the -print-multi-directory option:
$ sparc-rtems5-gcc -print-multi-directory -mcpu=leon3
leon3
It is crucial that the RTEMS BSP, support libraries and the application code are compiled consistently with a compatible set of machine options. Otherwise, in the best case errors during linking will occur or you may end up silently with undefined behaviour which results in sporadic run-time crashes. A wrong set of machine options may result in a running application, however, with degraded performance, e.g. hardware floating point unit is not used by the mathematical library.
For a list of architectures supported by RTEMS please have a look at the sections of the Board Support Packages chapter.
RTEMS executables are statically linked for a specific target therefore a precise and exact match can be made for the hardware that extracts the best possible performance. The compiler supports the variants to the instruction set and RTEMS extends the specialization to specific processors in an architecture. This specialization gives RTEMS a finer resolution of features and capabilities a specific device may offer allowing the kernel, drivers and application to make the most of those resources. The trade off is portability however this is not important because the executable are statically linked for a single target.
Note
RTEMS support dynamically load code through the dlopen interface. Loading code via this interface results in an executable image that is equivalent to statically linked executable of the same code. Dynamic loading is a system level tool for system architects.
6.3. Tiers¶
RTEMS has a tiered structure for architecture and BSPs. It provides:
A way to determine the state of a BSP in RTEMS.
A quaility measure for changes entering the RTEMS source code.
The RTEMS project supports RTEMS Architecture Tiers. Each architecture resided in one of the numbered tiers. The tiers are number 1 to 4 where Tier 1 is the highest tier and Tier 4 is the lowest. Architectures move between tiers based on the level of support and the level of testing that is performed. An architecture requires continual testing and reporting of test results to maintain a tier level. The RTEMS Project’s continuous integration testing program` continually monitors and reports the test results.
The RTEMS Architecture Tier system provides a defined way to determine the state of an architecture in RTEMS. Architectures age and support for them drops off and the RTEMS Project needs a way to determine if an architecture should stay and be supported or depreciated and removed. The tier system also provides users with a clear understanding of the state of an architecture in RTEMS, often useful when deciding on a processor for a new project. It can also let a user know the RTEMS Project needs support to maintain a specific architecture. Access to hardware to perform testing is a large and complex undertaking and the RTEMS Project is always looking for user support and help. If you can help please contact someone and let us know.
The tier structure in RTEMS is support by the Buildbot continuous integration server. Changes to RTEMS are automatically built and tested and the results indicate if a BSP currently meets its tier status. As the RTEMS Project does not own hardware for every BSP, it is critical that users provide test results on hardware of interest.
The rules for Tiers are:
A BSP can only be in one of the following tiers:
Tier
Description
1
The RTEMS Kernel must build without error.
Tests are run on target hardware.
2
The RTEMS Kernel must build without error.
Tests can be run on simulation.
3
The RTEMS Kernel must build without error.
There are no test results.
4
The RTEMS Kernel does not build.
5
The BSP is to be removed after the next release.
An architecuture’s tier is set by the highest BSP tier reached.
The tier level for a BSP is set by the RTEMS Project team. Movement of BSP between tier level requires agreement. The Buildbot results indicate the minimum current tier level.
Changes to RTEMS may result in a BSP not meeting its tier are acceptable if the change is accompanied by an announcement and a plan on how this is to be resolved. Temporary drops in tier are expected and should be brief.
Test results are set on a per BSP basis by the RTEMS Project team. Changes to the test result values requires agreement. The test results are defined as:
Passes
Expected Failures
Expected failures must be explicitly listed. A BSP is required to have a valid test result entry on target hardware to reach tier 1.
7. Board Support Packages¶
A Board Support Package or BSP is the software that glues a specific target or board or piece of hardware to RTEMS so it’s services are available to applications.
RTEMS contains a large number of BSPs for commonly available simulators and target hardware.
You can see the current BSP list in the RTEMS sources by asking RTEMS with:
$ ./rtems-bsps
7.1. aarch64 (AArch64)¶
There are no AArch64 BSPs yet.
7.2. arm (ARM)¶
7.2.1. altera-cyclone-v (Intel Cyclone V)¶
This BSP offers only one variant, the altcycv_devkit. This variant supports the Intel Cyclone V system on chip. The basic hardware initialization is not performed by the BSP. A boot loader with device tree support must be used to start the BSP, e.g. U-Boot.
The BSP is known to run on these boards:
7.2.1.1. Boot via U-Boot¶
The application executable file (ELF file) must be converted to an U-Boot image. Use the following commands:
arm-rtems5-objcopy -O binary app.exe app.bin
gzip -9 -f -c app.bin > app.bin.gz
mkimage -A arm -O linux -T kernel -a 0x00300000 -e 0x00300000 -n RTEMS -d app.bin.gz app.img
Use the following U-Boot commands to boot an application via TFTP download:
tftpboot ${loadaddr} app.img && run loadfdt && bootm ${loadaddr} - ${fdt_addr} ; reset
The loadfdt command may be not defined in your U-Boot environment. Just replace it with the appropriate commands to load the device tree at ${fdt_addr}.
7.2.1.2. Clock Driver¶
The clock driver uses the Cortex-A9 MPCore Global Timer.
7.2.1.3. Console Driver¶
The console driver supports up to two on-chip NS16550 UARTs. The console driver does not configure the pins.
7.2.1.4. I2C Driver¶
There is a legacy I2C driver. It should be converted to the I2C driver framework.
7.2.1.5. Network Interface Driver¶
The network interface driver is provided by the libbsd. It is initialized according to the device tree. It supports checksum offload.
7.2.1.6. MMC/SDCard Driver¶
The MMC/SDCard driver is provided by the libbsd. It is initialized according to the device tree. Pin re-configuration according to the serial clock frequency is not supported. DMA transfers are supported.
7.2.1.7. USB Host Driver¶
The USB host driver is provided by the libbsd. It is initialized according to the device tree. The driver works in polled mode.
7.2.1.8. Caveats¶
The clock and pin configuration support is quite rudimentary and mostly relies on the boot loader.
7.2.2. atsam¶
TODO.
7.2.3. beagle¶
This BSP supports four variants, beagleboardorig, beagleboardxm, beaglebonewhite and beagleboneblack. The basic hardware initialization is not performed by the BSP. A boot loader with device tree support must be used to start the BSP, e.g., U-Boot.
TODO(These drivers are present but not documented yet):
Clock driver.
Network Interface Driver.
SDcard driver.
GPIO Driver.
Console driver.
PWM Driver.
RTC driver.
7.2.3.1. Boot via U-Boot¶
To boot via uboot, the ELF must be converted to a U-Boot image like below:
arm-rtems5-objcopy hello.exe -O binary app.bin
gzip -9 app.bin
mkimage -A arm -O linux -T kernel -a 0x80000000 -e 0x80000000 -n RTEMS -d app.bin.gz rtems-app.img
7.2.3.2. Getting the Device Tree Blob¶
The Device Tree Blob (DTB) is needed to load the device tree while starting up the kernel. We build the dtb from the FreeBSD source matching the commit hash from the libbsd HEAD of freebsd-org. For example if the HEAD is at “19a6ceb89dbacf74697d493e48c388767126d418” Then the right Device Tree Source (DTS) file is: https://github.com/freebsd/freebsd/blob/19a6ceb89dbacf74697d493e48c388767126d418/sys/gnu/dts/arm/am335x-boneblack.dts
Please refer to the Device Tree to know more about building and applying the Device Trees.
7.2.3.3. Writing the uEnv.txt file¶
The uEnv.txt file is needed to set any environment variable before the kernel is loaded. Each line is a u-boot command that the uboot will execute during start up.
Add the following to a file named uEnv.txt:
setenv bootdelay 5
uenvcmd=run boot
boot=fatload mmc 0 0x80800000 rtems-app.img ; fatload mmc 0 0x88000000 am335x-boneblack.dtb ; bootm 0x80800000 - 0x88000000
7.2.3.4. I2C Driver¶
The Beagle has the i2c-0 device registered at initialization. For registering i2c-1 and i2c-2 bbb_register_i2c_1() and bbb_register_i2c_2() wrapper functions are respectively used.
For registering an I2C device with a custom path (say /dev/i2c-3) the function am335x_i2c_bus_register() has to be used.
The function prototype is given below:
int am335x_i2c_bus_register(
const char *bus_path,
uintptr_t register_base,
uint32_t input_clock,
rtems_vector_number irq
);
7.2.3.5. SPI Driver¶
The SPI device /dev/spi-0 can be registered with bbb_register_spi_0()
For registering with a custom path, the bsp_register_spi() can be used.
The function prototype is given below:
rtems_status_code bsp_register_spi(
const char *bus_path,
uintptr_t register_base,
rtems_vector_number irq
);
7.2.4. csb336¶
TODO.
7.2.5. edb7312¶
TODO.
7.2.6. gumstix¶
TODO.
7.2.7. imx (NXP i.MX)¶
This BSP offers only one variant, the imx7. This variant supports the i.MX 7Dual processor and the i.MX 6UL/ULL processor family (with slightly different clock settings). The basic hardware initialization is not performed by the BSP. A boot loader with device tree support must be used to start the BSP, e.g. U-Boot or barebox.
7.2.7.1. Build Configuration Options¶
The following options are available at the configure command line.
BSP_PRESS_KEY_FOR_RESETIf defined to a non-zero value, then print a message and wait until pressed before resetting board when application terminates.
BSP_RESET_BOARD_AT_EXITIf defined to a non-zero value, then reset the board when the application terminates.
BSP_PRINT_EXCEPTION_CONTEXTIf defined to a non-zero value, then print the exception context when an unexpected exception occurs.
BSP_FDT_BLOB_SIZE_MAXThe maximum size of the device tree blob in bytes (default is 262144).
CONSOLE_USE_INTERRUPTSUse interrupt driven mode for console devices (enabled by default).
IMX_CCM_IPG_HZThe IPG clock frequency in Hz (default is 67500000).
IMX_CCM_UART_HZThe UART clock frequency in Hz (default is 24000000).
IMX_CCM_ECSPI_HZThe ECSPI clock frequency in Hz (default is 67500000).
IMX_CCM_AHB_HZThe AHB clock frequency in Hz (default is 135000000).
IMX_CCM_SDHCI_HZThe SDHCI clock frequency in Hz (default is 196363000).
7.2.7.2. Clock settings for different boards¶
The default clock settings are targeted for an i.MX 7Dual evaluation board using U-Boot. Some other boards with different boot loaders need different settings:
Phytec phyCORE-i.MX 6ULL (system on module) with MCIMX6Y2CVM08AB and a barebox bootloader (version
2019.01.0-bsp-yocto-i.mx6ul-pd19.1.0):
IMX_CCM_IPG_HZ=66000000
IMX_CCM_UART_HZ=80000000
IMX_CCM_AHB_HZ=66000000
IMX_CCM_SDHCI_HZ=198000000
IMX_CCM_ECSPI_HZ=60000000
7.2.7.3. Boot via U-Boot¶
The application executable file (ELF file) must be converted to an U-Boot image. Use the following commands:
arm-rtems5-objcopy -O binary app.exe app.bin
gzip -9 -f -c app.bin > app.bin.gz
mkimage -A arm -O linux -T kernel -a 0x80200000 -e 0x80200000 -n RTEMS -d app.bin.gz app.img
Use the following U-Boot commands to boot an application via TFTP download:
tftpboot ${loadaddr} app.img && run loadfdt && bootm ${loadaddr} - ${fdt_addr} ; reset
The loadfdt command may be not defined in your U-Boot environment. Just replace it with the appropriate commands to load the device tree at ${fdt_addr}.
7.2.7.4. Boot via barebox¶
The same command like for U-Boot can be used to generate an application image. In a default configuration barebox expects an fdt image called oftree and a kernel image called zImage in the root folder of the bootable medium (e.g. an SD card).
7.2.7.5. Clock Driver¶
The clock driver uses the ARMv7-AR Generic Timer.
7.2.7.6. Console Driver¶
The console driver supports up to seven on-chip UARTs. They are initialized according to the device tree. The console driver does not configure the pins.
7.2.7.7. I2C Driver¶
I2C drivers are registered by the i2c_bus_register_imx() function. The I2C driver does not configure the pins.
#include <assert.h>
#include <bsp.h>
void i2c_init(void)
{
int rv;
rv = i2c_bus_register_imx("/dev/i2c-0", "i2c0");
assert(rv == 0);
}
7.2.7.8. SPI Driver¶
SPI drivers are registered by the spi_bus_register_imx() function. The SPI driver configures the pins according to the pinctrl-0 device tree property. SPI transfers with a continuous chip select are limited by the FIFO size of 64 bytes. The driver has no DMA support.
#include <assert.h>
#include <bsp.h>
void spi_init(void)
{
int rv;
rv = spi_bus_register_imx("/dev/spi-0", "spi0");
assert(rv == 0);
}
7.2.7.9. Network Interface Driver¶
The network interface driver is provided by the libbsd. It is initialized according to the device tree. It supports checksum offload and interrupt coalescing. IPv6 transmit checksum offload is not implemented. The interrupt coalescing uses the MII/GMII clocks and can be controlled by the following system controls:
dev.ffec.<unit>.int_coal.rx_time
dev.ffec.<unit>.int_coal.rx_count
dev.ffec.<unit>.int_coal.tx_time
dev.ffec.<unit>.int_coal.tx_count
A value of zero for the time or count disables the interrupt coalescing in the corresponding direction.
On the Phytec phyCORE-i.MX 6ULL modules the PHY needs an initialization for the clock. A special PHY driver handles that (ksz8091rnb). Add it to your libbsd config like that:
#define RTEMS_BSD_CONFIG_BSP_CONFIG
#define RTEMS_BSD_CONFIG_INIT
SYSINIT_DRIVER_REFERENCE(ksz8091rnb, miibus);
#include <machine/rtems-bsd-config.h>
7.2.7.10. MMC/SDCard Driver¶
The MMC/SDCard driver (uSDHC module) is provided by the libbsd. It is initialized according to the device tree. Pin re-configuration according to the serial clock frequency is not supported. Data transfers are extremely slow. This is probably due to the missing DMA support.
7.2.7.11. Caveats¶
The clock and pin configuration support is quite rudimentary and mostly relies on the boot loader. For a pin group configuration see imx_iomux_configure_pins(). There is no power management support.
7.2.8. lm3s69xx¶
TODO.
7.2.9. lpc176x¶
TODO.
7.2.10. imx (NXP i.MX)¶
This BSP offers only one variant, the imx7. This variant supports the i.MX 7Dual processor and the i.MX 6UL/ULL processor family (with slightly different clock settings). The basic hardware initialization is not performed by the BSP. A boot loader with device tree support must be used to start the BSP, e.g. U-Boot or barebox.
7.2.10.1. Build Configuration Options¶
The following options are available at the configure command line.
BSP_PRESS_KEY_FOR_RESETIf defined to a non-zero value, then print a message and wait until pressed before resetting board when application terminates.
BSP_RESET_BOARD_AT_EXITIf defined to a non-zero value, then reset the board when the application terminates.
BSP_PRINT_EXCEPTION_CONTEXTIf defined to a non-zero value, then print the exception context when an unexpected exception occurs.
BSP_FDT_BLOB_SIZE_MAXThe maximum size of the device tree blob in bytes (default is 262144).
CONSOLE_USE_INTERRUPTSUse interrupt driven mode for console devices (enabled by default).
IMX_CCM_IPG_HZThe IPG clock frequency in Hz (default is 67500000).
IMX_CCM_UART_HZThe UART clock frequency in Hz (default is 24000000).
IMX_CCM_ECSPI_HZThe ECSPI clock frequency in Hz (default is 67500000).
IMX_CCM_AHB_HZThe AHB clock frequency in Hz (default is 135000000).
IMX_CCM_SDHCI_HZThe SDHCI clock frequency in Hz (default is 196363000).
7.2.10.2. Clock settings for different boards¶
The default clock settings are targeted for an i.MX 7Dual evaluation board using U-Boot. Some other boards with different boot loaders need different settings:
Phytec phyCORE-i.MX 6ULL (system on module) with MCIMX6Y2CVM08AB and a barebox bootloader (version
2019.01.0-bsp-yocto-i.mx6ul-pd19.1.0):
IMX_CCM_IPG_HZ=66000000
IMX_CCM_UART_HZ=80000000
IMX_CCM_AHB_HZ=66000000
IMX_CCM_SDHCI_HZ=198000000
IMX_CCM_ECSPI_HZ=60000000
7.2.10.3. Boot via U-Boot¶
The application executable file (ELF file) must be converted to an U-Boot image. Use the following commands:
arm-rtems5-objcopy -O binary app.exe app.bin
gzip -9 -f -c app.bin > app.bin.gz
mkimage -A arm -O linux -T kernel -a 0x80200000 -e 0x80200000 -n RTEMS -d app.bin.gz app.img
Use the following U-Boot commands to boot an application via TFTP download:
tftpboot ${loadaddr} app.img && run loadfdt && bootm ${loadaddr} - ${fdt_addr} ; reset
The loadfdt command may be not defined in your U-Boot environment. Just replace it with the appropriate commands to load the device tree at ${fdt_addr}.
7.2.10.4. Boot via barebox¶
The same command like for U-Boot can be used to generate an application image. In a default configuration barebox expects an fdt image called oftree and a kernel image called zImage in the root folder of the bootable medium (e.g. an SD card).
7.2.10.5. Clock Driver¶
The clock driver uses the ARMv7-AR Generic Timer.
7.2.10.6. Console Driver¶
The console driver supports up to seven on-chip UARTs. They are initialized according to the device tree. The console driver does not configure the pins.
7.2.10.7. I2C Driver¶
I2C drivers are registered by the i2c_bus_register_imx() function. The I2C driver does not configure the pins.
#include <assert.h>
#include <bsp.h>
void i2c_init(void)
{
int rv;
rv = i2c_bus_register_imx("/dev/i2c-0", "i2c0");
assert(rv == 0);
}
7.2.10.8. SPI Driver¶
SPI drivers are registered by the spi_bus_register_imx() function. The SPI driver configures the pins according to the pinctrl-0 device tree property. SPI transfers with a continuous chip select are limited by the FIFO size of 64 bytes. The driver has no DMA support.
#include <assert.h>
#include <bsp.h>
void spi_init(void)
{
int rv;
rv = spi_bus_register_imx("/dev/spi-0", "spi0");
assert(rv == 0);
}
7.2.10.9. Network Interface Driver¶
The network interface driver is provided by the libbsd. It is initialized according to the device tree. It supports checksum offload and interrupt coalescing. IPv6 transmit checksum offload is not implemented. The interrupt coalescing uses the MII/GMII clocks and can be controlled by the following system controls:
dev.ffec.<unit>.int_coal.rx_time
dev.ffec.<unit>.int_coal.rx_count
dev.ffec.<unit>.int_coal.tx_time
dev.ffec.<unit>.int_coal.tx_count
A value of zero for the time or count disables the interrupt coalescing in the corresponding direction.
On the Phytec phyCORE-i.MX 6ULL modules the PHY needs an initialization for the clock. A special PHY driver handles that (ksz8091rnb). Add it to your libbsd config like that:
#define RTEMS_BSD_CONFIG_BSP_CONFIG
#define RTEMS_BSD_CONFIG_INIT
SYSINIT_DRIVER_REFERENCE(ksz8091rnb, miibus);
#include <machine/rtems-bsd-config.h>
7.2.10.10. MMC/SDCard Driver¶
The MMC/SDCard driver (uSDHC module) is provided by the libbsd. It is initialized according to the device tree. Pin re-configuration according to the serial clock frequency is not supported. Data transfers are extremely slow. This is probably due to the missing DMA support.
7.2.10.11. Caveats¶
The clock and pin configuration support is quite rudimentary and mostly relies on the boot loader. For a pin group configuration see imx_iomux_configure_pins(). There is no power management support.
7.2.11. raspberrypi¶
This BSP supports Raspberry Pi 1 and Raspberry Pi 2 currently. The support for Raspberry Pi 3 is work under progress. The default bootloader on the Raspberry Pi which is used to boot Raspbian or other OS can be also used to boot RTEMS. U-boot can also be used.
7.2.11.1. Setup SD card¶
The Raspberry Pis have an unconventional booting mechanism. The GPU boots first, initializes itself, runs the bootloader and starts the CPU. The bootloader looks for a kernel image, by default the kernel images must have a name of the form kernel*.img but this can be changed by adding kernel=<img_name> to config.txt.
You must provide the required files for the GPU to proceed. These files can be downloaded from the Raspberry Pi Firmware Repository. You can remove the kernel*.img files if you want too, but don’t touch the other files.
Copy these files in to a SD card with FAT filesystem.
7.2.11.2. Kernel image¶
The following steps show how to run hello.exe on a Raspberry Pi 2. The same instructions can be applied to Raspberry Pi 1 also. Other executables can be processed in a similar way.
To create the kernel image:
$ arm-rtems5-objcopy -Obinary hello.exe kernel.img
Copy the kernel image to the SD card.
Make sure you have these lines below, in your config.txt.
enable_uart=1
kernel_address=0x200000
kernel=kernel.img
7.2.11.3. Testing using QEMU¶
QEMU can be built using RSB. Navigate to <SOURCE_BUILDER_DIR>/rtems and run this command.
$ ../source-builder/sb-set-builder --prefix=<TOOLCHAIN_DIR> devel/qemu4.bset
Note: Replace <SOURCE_BUILDER_DIR> and <TOOLCHAIN_DIR> with the correct path of the directories. For example, if you used quick-start section as your reference, these two will be $HOME/quick-start/src/rsb and $HOME/quick-start/rtems/5 respectively,
QEMU along with GDB can be used for debugging, but it only supports Raspberry Pi 2 and the emulation is also incomplete. So some of the features might not work as expected.
Make sure your version of QEMU is newer than v2.6, because older ones don’t support Raspberry Pis.
$ qemu-system-arm -M raspi2 -m 1G -kernel hello.exe -serial mon:stdio -nographic -S -s
This starts QEMU and creates a socket at port localhost:1234 for GDB to connect.
The Device Tree Blob (DTB) is needed to load the device tree while starting up the kernel. The BSP uses information from this file to initialize the drivers.
Make sure you pass in the correct DTB file. There are currently two version of DTB for the Raspberry Pi 2 bcm2709-rpi-2-b.dtb and bcm2710-rpi-2-b.dtb. The bcm2709-rpi-2-b.dtb is for Raspberry Pi 2 Model B and bcm2710-rpi-2-b.dtb is for Raspberry Pi 2 Model B v1.2
We need to pass in the DTB file to GDB before running the example.
In a new terminal, run GDB using
$ arm-rtems5-gdb hello.exe
This will open GDB and will load the symbol table from hello.exe. Issue the following commands in the GDB prompt.
(gdb) tar remote:1234
(gdb) load
(gdb) restore bcm2709-rpi-2-b.dtb binary 0x2ef00000
(gdb) set $r2 = 0x2ef00000
This will connect GDB to QEMU and will load the DTB file and the application.
(gdb) continue
The continue command will run the executable.
Note: Add set scheduler-locking on in GDB if you have any issues running the examples.
7.2.12. realview-pbx-a9¶
TODO.
7.2.13. rtl22xx¶
TODO.
7.2.14. smdk2410¶
TODO.
7.2.15. tms570¶
TODO.
7.2.16. xen (Xen on ARM)¶
This BSP enables RTEMS to run as a guest virtual machine in AArch32 mode on the Xen hypervisor for ARMv8 platforms.
Drivers:
Clock: ARMv7-AR Generic Timer
Console: Virtual PL011 device
Interrupt: GICv2
BSP variants:
xen_virtual: completely virtualized guest with no dependence on underlying hardware
The xen_virtual BSP variant relies on standard Xen features, so it should be able to run on any ARMv8 platform.
Xen allows for the passthrough of hardware peripherals to guest virtual machines. BSPs could be added in the future targeting specific hardware platforms and include the appropriate drivers.
This BSP was tested with Xen running on the Xilinx Zynq UltraScale+ MPSoC using the Virtuosity distribution maintained by DornerWorks.
7.2.16.1. Execution¶
This procedure describes how to run the ticker sample application that should already be built with the BSP.
The ticker.exe file can be found in the BSP build tree at:
arm-rtems5/c/xen_virtual/testsuites/samples/ticker.exe
The ticker.exe elf file must be translated to a binary format.
arm-rtems5-objcopy -O binary ticker.exe ticker.bin
Then place the ticker.bin file on the dom0 filesystem.
From the dom0 console, create a configuration file ticker.cfg with the following contents.
name = "ticker"1G
kernel = "ticker.bin"
memory = 8
vcpus = 1
gic_version = "v2"
vuart = "sbsa_uart"
Create the virtual machine and attach to the virtual vpl011 console.
xl create ticker.cfg && xl console -t vuart ticker
To return back to the dom0 console, press both Ctrl and ] on your keyboard.
7.2.16.2. Additional Information¶
7.2.17. xilinx-zynq¶
TODO.
7.2.18. xilinx-zynqmp¶
This BSP supports the Xilinx Zynq UltraScale+ MPSoC platform.
7.7. m68k (Motorola 68000 / ColdFire)¶
7.7.1. av5282¶
TODO.
7.7.2. csb360¶
TODO.
7.7.3. gen68340¶
TODO.
7.7.4. gen68360¶
TODO.
7.7.5. genmcf548x¶
TODO.
7.7.6. mcf5206elite¶
TODO.
7.7.7. mcf52235¶
TODO.
7.7.8. mcf5225x¶
TODO.
7.7.9. mcf5235¶
TODO.
7.7.10. mcf5329¶
7.7.10.1. Overview¶
This BSP is heavily based on the MCF5235 BSP. The MCF5329EVB is a Motorola evaluation board (Zoom) with a LogicPD MCF5329-10 SODIMM-144 card. The development kit features the MCF5329 based Fire Engine, as well as a plug-in system-on-module containing 32 MB of DDR-SDRAM. The board also includes 2 MB of boot flash, 16 MB of NAND flash, a core frequency of 240MHz, an onboard 800x600 LCD controller, FEC, USB, uarts, CAN bus, QSPI, I2C, and 10/100 Ethernet.
You can find the link to MCF5329 Reference Manual below:
7.7.11. mrm332¶
TODO.
7.7.12. mvme147¶
TODO.
7.7.13. mvme147s¶
TODO.
7.7.14. mvme162¶
7.7.14.1. Overview¶
The MVME162 family provides OEMs and solution developers an ideal platform for embedded monitoring and control apllications it allows an OEM to minimize engineering expenses while integrating value-added hardware and software applications onto an off-the-shelf product. In order to provide the wide range of solutions, the MVME162 allows a variety of MPU, memory, and interface options such as floating-point, Ethernet, SCSI, and VME. The result is a variation of the MVME162 which most closely fits the application requirement.
There are a large number of model variations on this board. This was the first user submitted BSP and continues to be a fairly popular simply because at one point it was the highest selling VMEBus board of all time.
7.7.14.2. Board Setup¶
We will setup the RTEMS Lab Board initally to proceed further for the setup of TFTP transfer.
The env settings are:
MPU Clock Speed =25Mhz
162-Bug>env
Bug or System environment [B/S] = B?
Field Service Menu Enable [Y/N] = N?
Remote Start Method Switch [G/M/B/N] = B?
Probe System for Supported I/O Controllers [Y/N] = Y?
Negate VMEbus SYSFAIL* Always [Y/N] = N?
Local SCSI Bus Reset on Debugger Startup [Y/N] = N?
Local SCSI Bus Negotiations Type [A/S/N] = A?
Industry Pack Reset on Debugger Startup [Y/N] = Y?
Ignore CFGA Block on a Hard Disk Boot [Y/N] = Y?
Auto Boot Enable [Y/N] = N?
Auto Boot at power-up only [Y/N] = Y?
Auto Boot Controller LUN = 00?
Auto Boot Device LUN = 00?
Auto Boot Abort Delay = 15?
Auto Boot Default String [NULL for a empty string] = ?
ROM Boot Enable [Y/N] = N?
ROM Boot at power-up only [Y/N] = Y?
ROM Boot Enable search of VMEbus [Y/N] = N?
ROM Boot Abort Delay = 0?
ROM Boot Direct Starting Address = FF800000?
ROM Boot Direct Ending Address = FFDFFFFC?
Network Auto Boot Enable [Y/N] = N?
Network Auto Boot at power-up only [Y/N] = Y?
Network Auto Boot Controller LUN = 00?
Network Auto Boot Device LUN = 00?
Network Auto Boot Abort Delay = 5?
Network Auto Boot Configuration Parameters Pointer (NVRAM) = FFE0FF00?
Memory Search Starting Address = 00000000?
Memory Search Ending Address = 01000000?
Memory Search Increment Size = 00010000?
Memory Search Delay Enable [Y/N] = N?
Memory Search Delay Address = FFFFD20F?
Memory Size Enable [Y/N] = Y?
Memory Size Starting Address = 00000000?
Memory Size Ending Address = 01000000?
Base Address of Dynamic Memory = 00000000?
Size of Parity Memory = 00000000?
Size of ECC Memory Board #0 = 01000000?
Size of ECC Memory Board #1 = 00000000?
Base Address of Static Memory = FFE00000?
Size of Static Memory = 00020000?
Slave Enable #1 [Y/N] = Y?
Slave Starting Address #1 = 00000000?
Slave Ending Address #1 = 00FFFFFF?
Slave Address Translation Address #1 = 00000000?
Slave Address Translation Select #1 = 00000000?
Slave Control #1 = 03FF?
Slave Enable #2 [Y/N] = N?
Slave Starting Address #2 = 00000000?
Slave Ending Address #2 = 00000000?
Slave Address Translation Address #2 = 00000000?
Slave Address Translation Select #2 = 00000000?
Slave Control #2 = 0000?
Master Enable #1 [Y/N] = Y?
Master Starting Address #1 = 01000000?
Master Ending Address #1 = EFFFFFFF?
Master Control #1 = 0D?
Master Enable #2 [Y/N] = N?
Master Starting Address #2 = 00000000?
Master Ending Address #2 = 00000000?
Master Control #2 = 00?
Master Enable #3 [Y/N] = N?
Master Starting Address #3 = 00000000?
Master Ending Address #3 = 00000000?
Master Control #3 = 00?
Master Enable #4 [Y/N] = N?
Master Starting Address #4 = 00000000?
Master Ending Address #4 = 00000000?
Master Address Translation Address #4 = 00000000?
Master Address Translation Select #4 = 00000000?
Master Control #4 = 00?
Short I/O (VMEbus A16) Enable [Y/N] = Y?
Short I/O (VMEbus A16) Control = 01?
F-Page (VMEbus A24) Enable [Y/N] = Y?
F-Page (VMEbus A24) Control = 02?
ROM Access Time Code = 03?
FLASH Access Time Code = 02?
MCC Vector Base = 05?
VMEC2 Vector Base #1 = 06?
VMEC2 Vector Base #2 = 07?
VMEC2 GCSR Group Base Address = D2?
VMEC2 GCSR Board Base Address = 00?
VMEbus Global Time Out Code = 01?
Local Bus Time Out Code = 02?
VMEbus Access Time Out Code = 02?
IP A Base Address = 00000000?
IP B Base Address = 00000000?
IP C Base Address = 00000000?
IP D Base Address = 00000000?
IP D/C/B/A Memory Size = 00000000?
IP D/C/B/A General Control = 00000000?
IP D/C/B/A Interrupt 0 Control = 00000000?
IP D/C/B/A Interrupt 1 Control = 00000000?
To setup the Server/Client IP Addresses for the TFTP Transfer, we will use the NIOT command. NIOT (Network I/O Teach) is a 162-Bug’s debugger command commonly used to setup the Server/Client IP Addresses for the TFTP Transfer.
The NIOT command goes something like this:
162-Bug>niot
Controller LUN =00?
Device LUN =00?
Node Control Memory Address =FFE10000?
Client IP Address =192.168.1.245?
Server IP Address =192.168.1.92?
Subnet IP Address Mask =255.255.255.0?
Broadcast IP Address =192.168.1.255?
Gateway IP Address =0.0.0.0?
Boot File Name ("NULL" for None) =/mvme162.img?
Argument File Name ("NULL" for None) =?
Boot File Load Address =00020000?
Boot File Execution Address =00020000?
Boot File Execution Delay =00000000?
Boot File Length =00000000?
Boot File Byte Offset =00000000?
BOOTP/RARP Request Retry =00?
TFTP/ARP Request Retry =00?
Trace Character Buffer Address =00000000?
BOOTP/RARP Request Control: Always/When-Needed (A/W)=A?
BOOTP/RARP Reply Update Control: Yes/No (Y/N) =Y?
7.7.14.3. Downloading and Executing¶
Download from the TFTP server using the 162-Bug’s “NBO” (Network Boot Operating System) command:
162-Bug>nbo
Network Booting from: VME162, Controller 0, Device 0
Loading: /mvme162.img
Client IP Address = 192.168.1.245
Server IP Address = 192.168.1.92
Gateway IP Address = 0.0.0.0
Subnet IP Address Mask = 255.255.255.0
Boot File Name = /mvme162.img
Argument File Name =
Network Boot File load in progress... To abort hit <BREAK>
Bytes Received =&356528, Bytes Loaded =&356528
Bytes/Second =&89132, Elapsed Time =4 Second(s)
The program will automatically run when download is complete.
7.7.15. mvme167¶
TODO.
7.7.16. uC5282¶
TODO.
7.8. microblaze (Microblaze)¶
There are no Microblaze BSPs yet.
7.9. mips (MIPS)¶
7.9.1. csb350¶
TODO.
7.9.2. hurricane¶
TODO.
7.9.3. jmr3904¶
TODO.
7.9.4. malta¶
TODO.
7.9.5. rbtx4925¶
TODO.
7.9.6. rbtx4938¶
TODO.
7.13. powerpc (PowerPC)¶
7.13.1. beatnik¶
TODO.
7.13.2. gen5200¶
TODO.
7.13.3. gen83xx¶
TODO.
7.13.4. haleakala¶
TODO.
7.13.5. motorola_powerpc¶
7.13.5.1. Boot Image Generation¶
The application executable file (ELF file) must be converted to a boot image. Use the following commands:
powerpc-rtems5-objcopy -O binary -R .comment -S ticker.exe rtems
gzip -9 -f rtems
powerpc-rtems5-ld -o ticker.boot bootloader.o --just-symbols=ticker.exe -b binary rtems.gz -T ppcboot.lds -no-warn-mismatch
powerpc-rtems5-objcopy -O binary ticker.boot ticker.bin
7.13.6. mpc55xxevb¶
TODO.
7.13.7. mpc8260ads¶
TODO.
7.13.8. mvme3100¶
TODO.
7.13.9. mvme5500¶
TODO.
7.13.10. psim¶
TODO.
7.13.11. qemuppc¶
TODO.
7.13.12. qoriq (QorIQ)¶
The BSP for the QorIQ chip family offers three variants. The qoriq_e500 variant supports the P-series chips such as P1020, P2010 and P2020. The qoriq_e6500_32 (32-bit ISA) and qoriq_e6500_64 (64-bit ISA) variants support the T-series chips such as T2080 and T4240. The basic hardware initialization is not performed by the BSP. A boot loader with device tree support must be used to start the BSP, e.g. U-Boot.
The BSP is known to run on these boards:
NXP P1020RDB
MicroSys miriac MPX2020 (System on Module)
Artesyn MVME2500 (VME64x SBC)
NXP T2080RDB
NXP T4240RDB
MEN G52A (CompactPCI Serial)
The qoriq_core_0 and qoriq_core_1 variants should be used with care. They are inteded for a RTEMS_MULTIPROCESSING configuration on the P1020.
7.13.12.1. Boot via U-Boot¶
The application executable file (ELF file) must be converted to an U-Boot image. Use the following commands:
powerpc-rtems5-objcopy -O binary app.exe app.bin
gzip -9 -f -c app.bin > app.bin.gz
mkimage -A ppc -O linux -T kernel -a 0x4000 -e 0x4000 -n RTEMS -d app.bin.gz app.img
Use the following U-Boot commands to boot an application via TFTP download:
tftpboot ${loadaddr} app.img && run loadfdt && bootm ${loadaddr} - ${fdt_addr} ; reset
7.13.12.2. Clock Driver¶
The clock driver uses two MPIC global timer (QORIQ_CLOCK_TIMER and QORIQ_CLOCK_TIMECOUNTER). In case QORIQ_IS_HYPERVISOR_GUEST is defined, then the PowerPC decrementer is used.
7.13.12.3. Console Driver¶
The console driver supports the on-chip NS16550 compatible UARTs. In case QORIQ_IS_HYPERVISOR_GUEST is defined, then the EPAPR byte channel is used for the console device.
7.13.12.4. Network Interface Driver¶
The network interface driver is provided by the libbsd. The DPAA is supported including 10Gbit/s Ethernet.
7.13.12.5. Topaz Hypervisor Guest¶
For a Topaz hypervisor guest configuration use:
../configure --enable-rtemsbsp=qoriq_e6500_32 \
QORIQ_IS_HYPERVISOR_GUEST=1 \
QORIQ_UART_0_ENABLE=0 \
QORIQ_UART_1_ENABLE=0 \
QORIQ_TLB1_ENTRY_COUNT=16
You may have to adjust the linker command file according to your partition configuration.
7.13.13. ss555¶
TODO.
7.13.14. t32mppc¶
TODO.
7.13.15. tqm8xx¶
TODO.
7.13.16. virtex¶
TODO.
7.13.17. virtex4¶
TODO.
7.13.18. virtex5¶
TODO.
7.14. riscv (RISC-V)¶
7.14.1. riscv¶
This BSP offers 13 variants:
rv32i
rv32iac
rv32im
rv32imac
rv32imafc
rv32imafd
rv32imafdc
rv64imac
rv64imac_medany
rv64imafd
rv64imafd_medany
rv64imafdc
rv64imafdc_medany
frdme310arty
Each variant corresponds to a GCC multilib. A particular variant reflects an ISA with ABI and code model choice.
The basic hardware initialization is not performed by the BSP. A boot loader with device tree support must be used to start the BSP, e.g. BBL. The BSP must be started im machine mode.
The reference platform for this BSP is the Qemu virt machine.
7.14.1.1. Build Configuration Options¶
The following options are available at the configure command line.
BSP_PRESS_KEY_FOR_RESETIf defined to a non-zero value, then print a message and wait until pressed before resetting board when application terminates.
BSP_RESET_BOARD_AT_EXITIf defined to a non-zero value, then reset the board when the application terminates.
BSP_PRINT_EXCEPTION_CONTEXTIf defined to a non-zero value, then print the exception context when an unexpected exception occurs.
BSP_FDT_BLOB_SIZE_MAXThe maximum size of the device tree blob in bytes (default is 65536).
BSP_CONSOLE_BAUDThe default baud for console driver devices (default 115200).
RISCV_MAXIMUM_EXTERNAL_INTERRUPTSThe maximum number of external interrupts supported by the BSP (default 64).
RISCV_ENABLE_HTIF_SUPPORTEnables the HTIF support if defined to a non-zero value, otherwise it is disabled (disabled by default).
RISCV_CONSOLE_MAX_NS16550_DEVICESThe maximum number of NS16550 devices supported by the console driver (2 by default).
RISCV_RAM_REGION_BEGINThe begin of the RAM region for linker command file (default is 0x70000000 for 64-bit with -mcmodel=medlow and 0x80000000 for all other).
RISCV_RAM_REGION_SIZEThe size of the RAM region for linker command file (default 64MiB).
RISCV_ENABLE_FRDME310ARTY_SUPPORTEnables support sifive Freedom E310 Arty board if defined to a non-zero value,otherwise it is disabled (disabled by default)
7.14.1.2. Interrupt Controller¶
Exactly one Core Local Interruptor (CLINT) and exactly one Platform-Level Interrupt Controller (PLIC) are supported. The maximum number of external interrupts supported by the BSP is defined by the RISCV_MAXIMUM_EXTERNAL_INTERRUPTS BSP option.
7.14.1.3. Clock Driver¶
The clock driver uses the CLINT timer.
7.14.1.4. Console Driver¶
The console driver supports devices compatible to
“ucb,htif0” (depending on the
RISCV_ENABLE_HTIF_SUPPORTBSP option),“ns16550a” (see
RISCV_CONSOLE_MAX_NS16550_DEVICESBSP option), and“ns16750” (see
RISCV_CONSOLE_MAX_NS16550_DEVICESBSP option).“sifive,uart0” (see
RISCV_ENABLE_FRDME310ARTY_SUPPORTBSP option).
They are initialized according to the device tree. The console driver does not configure the pins or peripheral clocks. The console device is selected according to the device tree “/chosen/stdout-path” property value.
7.15. sh (SuperH)¶
7.15.1. gensh1¶
TODO.
7.15.2. gensh2¶
TODO.
7.15.3. gensh4¶
TODO.
7.15.4. shsim¶
TODO.
7.19. x86_64¶
7.19.1. amd64¶
This BSP offers only one variant, amd64. The BSP can run on UEFI-capable systems by using FreeBSD’s bootloader, which then loads the RTEMS executable (an ELF image).
Currently only the console driver and context initialization and switching are functional (to a bare minimum), but this is enough to run the hello.exe sample in the RTEMS testsuite.
7.19.1.1. Build Configuration Options¶
There are no options available to configure at build time, at the moment.
7.19.1.2. Testing with QEMU¶
To test with QEMU, we need to:
Build / install QEMU (most distributions should have it available on the package manager).
Build UEFI firmware that QEMU can use to simulate an x86-64 system capable of booting a UEFI-aware kernel, through the
--biosflag.
7.19.1.2.1. Building TianoCore’s UEFI firmware, OVMF¶
Complete detailed instructions are available at TianoCore’s Github’s wiki.
Quick instructions (which may fall out of date) are:
$ git clone git://github.com/tianocore/edk2.git
$ cd edk2
$ make -C BaseTools
$ . edksetup.sh
Then edit Conf/target.txt to set:
ACTIVE_PLATFORM = OvmfPkg/OvmfPkgX64.dsc
TARGET = DEBUG
TARGET_ARCH = X64
# You can use GCC46 as well, if you'd prefer
TOOL_CHAIN_TAG = GCC5
Then run build in the edk2 directory - the output should list the location of the OVMF.fd file, which can be used with QEMU to boot into a UEFI shell.
You can find the OVMF.fd file like this as well in the edk2 directory:
$ find . -name "*.fd"
./Build/OvmfX64/DEBUG_GCC5/FV/MEMFD.fd
./Build/OvmfX64/DEBUG_GCC5/FV/OVMF.fd # the file we're looking for
./Build/OvmfX64/DEBUG_GCC5/FV/OVMF_CODE.fd
./Build/OvmfX64/DEBUG_GCC5/FV/OVMF_VARS.fd
7.19.1.3. Boot RTEMS via FreeBSD’s bootloader¶
The RTEMS executable produced (an ELF file) needs to be placed in the FreeBSD’s /boot/kernel/kernel’s place.
To do that, we first need a hard-disk image with FreeBSD installed on it. Download FreeBSD’s installer “memstick” image for amd64 and then run the following commands, replacing paths as appropriate.
$ qemu-img create freebsd.img 8G
$ OVMF_LOCATION=/path/to/ovmf/OVMF.fd
$ FREEBSD_MEMSTICK=/path/to/FreeBSD-11.2-amd64-memstick.img
$ qemu-system-x86_64 -m 1024 -serial stdio --bios $OVMF_LOCATION \
-drive format=raw,file=freebsd.img \
-drive format=raw,file=$FREEBSD_MEMSTICK
The first time you do this, continue through and install FreeBSD. FreeBSD’s installation guide may prove useful if required.
Once installed, build your RTEMS executable (an ELF file), for eg. hello.exe. We need to transfer this executable into freebsd.img’s filesystem, at either /boot/kernel/kernel or /boot/kernel.old/kernel (or elsewhere, if you don’t mind user FreeBSD’s loader’s prompt to boot your custom kernel).
If your host system supports mounting UFS filesystems as read-write (eg. FreeBSD), go ahead and:
Mount
freebsd.imgas read-writeWithin the filesystem, back the existing FreeBSD kernel up (i.e. effectively
cp -r /boot/kernel /boot/kernel.old).Place your RTEMS executable at
/boot/kernel/kernel
If your host doesn’t support mounting UFS filesystems (eg. most Linux kernels), do something to the effect of the following.
On the host
# Upload hello.exe anywhere accessible within the host
$ curl --upload-file hello.exe https://transfer.sh/rtems
Then on the guest (FreeBSD), login with root and
# Back the FreeBSD kernel up
$ cp -r /boot/kernel/ /boot/kernel.old
# Bring networking online if it isn't already
$ dhclient em0
# You may need to add the --no-verify-peer depending on your server
$ fetch https://host.com/path/to/rtems/hello.exe
# Replace default kernel
$ cp hello.exe /boot/kernel/kernel
$ reboot
After rebooting, the RTEMS kernel should run after the UEFI firmware and FreeBSD’s bootloader. The -serial stdio QEMU flag will let the RTEMS console send its output to the host’s stdio stream.
7.19.1.4. Paging¶
During the BSP’s initialization, the paging tables are setup to identity-map the first 512GiB, i.e. virtual addresses are the same as physical addresses for the first 512GiB.
The page structures are set up statically with 1GiB super-pages.
Note
Page-faults are not handled.
Warning
RAM size is not detected dynamically and defaults to 1GiB, if the configuration-time RamSize parameter is not used.
7.19.1.5. Interrupt Setup¶
Interrupt vectors 0 through 32 (i.e. 33 interrupt vectors in total) are setup as “RTEMS interrupts”, which can be hooked through rtems_interrupt_handler_install.
The Interrupt Descriptor Table supports a total of 256 possible vectors (0 through 255), which leaves a lot of room for “raw interrupts”, which can be hooked through _CPU_ISR_install_raw_handler.
Since the APIC needs to be used for the clock driver, the PIC is remapped (IRQ0 of the PIC is redirected to vector 32, and so on), and then all interrupts are masked to disable the PIC. In this state, the PIC may _still_ produce spurious interrupts (IRQ7 and IRQ15, redirected to vector 39 and vector 47 respectively).
The clock driver triggers the initialization of the APIC and then the APIC timer.
The I/O APIC is not supported at the moment.
Note
IRQ32 is reserved by default for the APIC timer (see following section).
IRQ255 is reserved by default for the APIC’s spurious vector.
Warning
Besides the first 33 vectors (0 through 32), and vector 255 (the APIC spurious vector), no other handlers are attached by default.
7.19.1.6. Clock Driver¶
The clock driver currently uses the APIC timer. Since the APIC timer runs at the CPU bus frequency, which can’t be detected easily, the PIT is used to calibrate the APIC timer, and then the APIC timer is enabled in periodic mode, with the initial counter setup such that interrupts fire at the same frequency as the clock tick frequency, as requested by CONFIGURE_MICROSECONDS_PER_TICK.
7.19.1.7. Console Driver¶
The console driver defaults to using the COM1 UART port (at I/O port 0x3F8), using the NS16550 polled driver.
8. Executables¶
This section discusses what an RTEMS executable is and what happens when you execute it in a target. The section discusses how an application executable is created, what happens when an executable is loaded and run, debugging an execiutable, and creating and dynamically loading code.
8.1. RTEMS Executable¶
Running executables is the most important part of working with RTEMS, it is after all how you run your application and use the RTEMS kernel services.
An RTEMS executable is embedded in a target and executing an embedded executable has challenges not faced when executing software on a desktop or server computer. A desktop or server operating system kernel provides all the support needed to bring an executable’s code and data into a process’s address space passing control to it and cleaning up when it exits. An embedded target has to provide similar functionality to execute an embedded executable.
An RTEMS Source Builder (RSB) built RTEMS tool chain is used to create RTEMS executables. The tool chain executable creates a fixed position statically linked Extendable Loader Format (ELF) file that contains the RTEMS kernel, standard libraries, third-party libraries and application code. RTEMS executes in a single address space which means it does not support the fork or exec system calls so statically linking all the code is the easiest and best way to create an executable.
An RTEMS application is constructed vertically with the RTEMS kernel, BSP support code and drivers close to the hardware, above which sit the RTEMS Application Programming Interfaces (API) for control of threads, mutex and other resources an application may use. Middle-ware services like networking, interpreted languages, and protocol stacks sit between the RTEMS APIs and the application components. The software built into an executable can be see as a vertical software stack.
Vertical Software Stack¶
8.2. Building an Application¶
RTEMS views any code it is running and using it’s interfaces as an application. RTEMS conforms to a number of international standards such as POSIX and can build and run portable code written in languages such as C, C++ and Ada.
Applications are built from source into ELF object files, third-party packages can be built as libraries or they can be imported as source into an application code base. The application, third-party packages, RTEMS and standard libraries are linked to create the RTEMS executable. The executable is transferred to the target and a bootloader loads it from the non-volatile storage into RAM or the code is executed in place in the non-volatile storage. The target hardware defines what happens.
Building an Application¶
The standard and third-party libraries are a collection of object files built using the same set of tools the application source is compiled with. The package collects it’s object files into an archive or library.
RTEMS does not provide a standard application build system. The RTEMS ecosystem provides support so a range of build systems can be used. Applications can be built with make, autotools, cmake, waf and more. User should select a build system that meets their project, system, corporate or personal needs.
8.2.1. Machine Flags and ABI¶
All code in an RTEMS executable must be built with the same machine flags. The machine flags control the instruction set and application binary interface (ABI) the compiler generates. As the executable is statically linked all code must use the same instruction set the hardware is configured to support and all code must conform to the same ABI. Any variation can result in unpredictable behavior such as crashes, failures or lock ups. It is recommend an executable is built with the same or equivalent tool set. Mixing of tool set versions can also result in undefined behavior. The RTEMS tool rtems-execinfo can audit an RTEMS executable and list the machine flags and compilers used.
RTEMS by default does not support instruction emulation for unsupported instructions. RTEMS applications are normally built from source so binary compatibility is not as important as performance. Instruction emulation is costly to execute and rebuilding the executable with the correct instruction set only needs to be done once.
8.3. Target Execution¶
Fixed position statically linked executables have a fixed address in a target’s address space. The location in the address space for code, data and read-only data is fixed. The BSP defines the memory map and it is set by the BSP developer based on the target’s hardware requirements and it’s bootloader.
Targets typically contains a bootloader that is executed after the target’s processor exits reset. A bootloader is specific to a target’s processor and hardware configuration and is responsible for the low level initialization of the hardware resources needed to load and execute an operating system’s kernel. In the case of RTEMS this is the RTEMS executable.
Bootloaders vary in size, complexity and functionality. Some architectures have a number of bootloader stages and others have only minimal support. An example of a high end system is Xilinx’s Zynq processor with three stages. First a mask ROM in the System On Chip (SOC) executes after reset loading a first stage bootloader (FSBL) from an SD card, QSPI flash or NAND flash depending on signals connected to the device. The FSBL loads a second stage bootloader (SSBL) such as U-Boot and this loads the kernel. U-Boot can be configured to load a kernel from a range of media and file system formats as well as over a network using a number of protocols. This structure provides flexibility at the system level to support development environments such as a workshop or laboratory through to tightly control production configurations.
Bootloaders often have custom formats for the executable image they load. The formats can be simple to keep the bootloader simple or complex to support check-sums, encryption or redundancy in case an image becomes corrupted. A bootloader often provides a host tool that creates the required file from the RTEMS executable’s ELF file.
If RTEMS is to run from RAM the bootloader reads the image and loads the code, initialized data and read-only data into the RAM and then jumps to a known entry point. If the code is executed from non-volatile storage the process to write the image into that storage will have extracted the various binary parts and written those to the correct location.
The important point to note is the binary parts of the executable are somehow loaded into the target’s address space ready to execute. The way this done may vary but the out come is always the same, the binary code, data and read-only data is resident in the processor’s address space at the BSP defined addresses.
8.4. BSP Initialization¶
The bootloader jumps or calls the RTEMS executable’s entry point, normally a fixed address. The BSP entry point or start up code performs:
Low level processor specific initialization that such as setting control registers so the processor is operating in a mode RTEMS is built for
Cache flushing, clearing and invalidation
Memory management unit (MMU) set up if required
Clear the uninitialized data section
Process a command line if supported by the bootloader
Call
bootcardwhich disabled interrupts, saves away a command line if the BSP supports it then call the RTEMS kernel early initialize entry pointrtems_initialize_executive. This call never returns.
Further BSP initialization happens as part of RTEMS kernel’s System Initialization process. The following handlers are declared and if provided are placed at the beginning of the initialization handler list. The BSP can provides:
bsp_work_area_initializeThis function determines the amount of memory that can be given to RTEMS for the workspace and the C library heap which
mallocuses. The call typically uses thebsp_work_area_initialize_defaultto perform actually perform the initialization.bsp_startThis function is specialized for each architecture and even for some BSPs. It performs the low level initialization RTEMS needs so it can run on the architecture and BSP.
bsp_predriver_hookThis function can be used to initialize hardware drivers depend on such as configuring an interrupt controller. The default version is empty and does nothing.
BSPs all perform similar operations with common functionality and the RTEMS kernel provides common code that can be shared between BSPs. The use of the common code is encouraged for all new BSPs.
8.5. RTEMS Initialization¶
The RTEMS kernel initialization is:
Invoke the registered system initialization handlers
Set the system state to up
If the kernel supports SMP request multitasking start. All online cores are transferred to the ready to start multitasking state.
Start threaded multitasking. RTEMS starts multitasking by getting the first thread to run and dispatching it.
C++ static object constructors are called in the context of the first running thread before the thread body is entered.
8.5.1. System Initialization Handlers¶
RTEMS supports the automatic registration of services used in applications. This method of initialization automatically configures RTEMS with only the services used in an application. There is no manual configuration of services used and no updating of initialization function tables.
RTEMS uses specialized sections in the ELF executable to perform this task. The system is based on the FreeBSD SYSINT Framework. Ordered initialization is performed before multitasking is started.
The RTEMS Tool rtems-exeinfo can provide some detail about the registered handlers. The following shows the initialization handlers for the Hello World sample application in the RTEMS kernel’s testsuite:
.. code-block:: none
$ rtems-exeinfo –init arm-rtems5/c/xilinx_zynq_zedboard/testsuites/samples/hello.exe RTEMS Executable Info 5.5416cfa39dd6 $ rtems-exeinfo –init arm-rtems5/c/xilinx_zynq_zedboard/testsuites/samples/hello.exe exe: arm-rtems5/c/xilinx_zynq_zedboard/testsuites/samples/hello.exe
- Compilation:
- Producers: 2
- Common flags: 4
- Init sections: 2
- .init_array
0x001047c1 frame_dummy
- .rtemsroset
0x00104c05 bsp_work_area_initialize 0x00104c41 bsp_start 0x0010eb45 zynq_debug_console_init 0x0010ec19 rtems_counter_sysinit 0x0010b779 _User_extensions_Handler_initialization 0x0010c66d rtems_initialize_data_structures 0x00107751 _RTEMS_tasks_Manager_initialization 0x0010d4f5 _POSIX_Keys_Manager_initialization 0x0010dd09 _Thread_Create_idle 0x0010cf01 rtems_libio_init 0x001053a5 rtems_filesystem_initialize 0x0010546d _Console_simple_Initialize 0x0010c715 _IO_Initialize_all_drivers 0x001076d5 _RTEMS_tasks_Initialize_user_tasks_body 0x0010cfa9 rtems_libio_post_driver
The section .rtemsroset lists the handlers called in order. The handlers can be split into the BSP initialization handlers that start the BSP:
bsp_work_area_initializebsp_startzynq_debug_console_initrtems_counter_sysinit
And the remainder are handlers for services used by the application. The list varies based on the services the application uses.
8.6. Debugging¶
An RTEMS executable is debugged by loading the code, data and read-only data into a target with a debugger connected. The debugger running on a host computer accesses the ELF file reading the debug information it contains.
The executable being debugged needs to be built with the compiler and linker debug options enabled. Debug information makes the ELF executable file large but it does not change the binary footprint of the executable when resident in the target. Target boot loaders and file conversion tools extract the binary code, data and read-only data to create the file embedded on the target.
An ELF executable built with debug information contains DWARF debug information. DWARF is a detailed description of the executable a debugger uses to locate functions, find data, understand the type and structure of a variable, and know how much entry code every call has. The debugger uses this information to set breaks points, step functions, step instructions, view the data and much more.
We recommend the compiler and linker debug options are always enabled. An ELF file with debug information can be used to investigate a crash report from a production system if the production ELF image is archived. The RTEMS tools chain provides tools that can take an address from a crash dump and find the corresponding instruction and source line. The extra size the debug information adds does not effect the target footprint and the extra size on a host is small compared to the benefits it brings.
A desktop or server operating system’s kernel hosts the executable being debugged handling the interaction with the executable and the debugger. The debugger knows how to communicate to the kernel to get the information it needs. Debugging an embedded executable needs an extra piece called an agent to connect the target to the debugger. The agent provides a standard remote interface to the debugger and an agent specific connection to the target.
Embedded Executable Debugging¶
The RTEMS tool chain provides the GNU debugger GDB. GDB has a remote protocol that can run over networks using TCP and UDP protocols. The GDB remote protocol is available in a number of open source and commercial debugging solutions. Network debugging using the remote protocol helps setting up a laboratory, the targets can be remote from the developers desktop allowing for better control of the target hardware while avoiding the need to plug devices in to an expensive desktop or server machine.
The following are some examples of GDB and GDB server environments RTEMS supports.
QEMU contains a debugging agent for the target being simulated. A QEMU command line option enables a GDB server and the simulator manages the interaction with the target processor and it’s memory and caches.
QEMU Executable Debugging¶
OpenOCD is a JTAG debugging package that interfaces to a wide of JTAG pods. JTAG is a low level high speed serial interface modern processors provide as a means of controlling the core processing logic. The features available depend on the architecture and processor. Typical functions include:
Processor control and register access
System level register access to allow SOC initialization
General address space access
Cache and MMU control
Break and watch points
OpenOCD JTAG Executable Debugging¶
The RTEMS kernel has a debugging agent called libdebugger. This is a software based agent that runs within RTEMS using network services to provide a remote GDB protocol interface. A growing number of architectures are supported. The RTEMS debugging agent is for application development providing thread aware stop model debug experience.
Libdebugger Executable Debugging¶
8.7. Dynamic Loader¶
RTEMS supports dynamically loading of executable code and data in the form of object files into a running system where the run-time loaded code can be executed and data accessed
This section describes RTEMS loader, preparing and loading executable code into a running system, the supported architectures and any limitation that may exist with an architecture.
The RTEMS operating system contains a link editor that runs on the target. The link editor supports loading Extendable Linker Format (ELF) relocatable executable object files locating the code and data in the target’s address space as it is loaded. An executable object file’s external references to function identifiers and data object identifiers are resolved and any external symbols can be made available in the global symbol table. The executing performance of dynamically loaded code is similar to the same code statically linked into an executable. This is a core requirement of the RTEMS link editor.
Run Time Loader (libdl)¶
The RTEMS operating system’s dynamic loader is not the same as the dynamic shared library support Unix or Windows have. Those operating systems use dynamic loading to share code between processes and this is an important feature in their design. RTEMS is a single address space operating system and that means there is no ability to share code at run-time. As a result code is loaded in a similar manner to static linking removing the need for any overheads sharing code may have.
To load an executable object file it must be resident on a target and accessible by RTEMS’s file system. The executable object file can be a single file or a collection in a library stored using the Unix standard archive format. The RTEMS loader supports the extended GNU format for long file names in archives.
The RTEMS developers do not see dynamically loading of code as a real-time activity. A system should not respond to real-time external events by loading code. The loading of code should happen before a system is considered available and the activity the system is experiencing is low and stable.
The statically linked executable that is loaded and run after reset is called the base image. The base image contains your base application that is used to dynamically load code, a global symbol table, the parts of the RTEMS operating system code used in the base image as well as functions and data from the tool suite libraries and packages you are using. Only the software referenced is used to create the base image. The parts of the libraries not referenced are not part of the executable or present in the global symbol table.
Application software can locate a symbol by name and call the address or reference the data at that address. A function identifier located by a symbol does not have it’s signatures checked, it is the responsibility of the caller to make sure the function is called with the correct arguments. It is the same for data objects, there is no type checking. Symbol versioning is not supported and supporting it does not make sense within the RTEMS operating system. An RTEMS target system is closed to normal users and software needs to be built from the same tool set and header files used to the build the base image.
An executable object file’s text or code has to be built for the target’s architecture it is loaded on and it must be built with the same ABI flags the base image is built with. See Machine Flags and ABI.
8.7.1. System Design¶
The use of dynamic loading in a project is a system design decision. Some systems will have strict requirements where loading code into a live system is not allowed while other projects will benefit from the system level flexibility dynamically loading code provides.
Code loaded at run time needs to be resident or accessible to the target via RTEMS’s file system. Targets that have suitable media or a network interface to NFS servers to hold the executable object and library files are best suited.
Dynamically loading code uses more memory than statically linking the same code into the base image. The link editor maintains symbol tables where each symbol is a string, an address, and some additional data. The executable object files resident in memory each have data to manage them, the memory they use, and any dependencies they might have. The link editor is designed to minimize the memory overheads however only statically linked executables have no memory overhead.
The link editor relocates the code and data into RAM fixing it to the load address as it is loaded. A target needs to have suitably configured memory available for the executable object file to reside in. The memory must be able to support read, write and executable type access. Fine control of the memory and it’s modes can be supported using a customer allocator. Examples are systems that have a custom memory map, specialized memory for the execution of code or a requirement for read-only executable sections.
The load address of an executable object file is determined by the load order and the allocator used. The default allocator for the link editor is the system heap which means the location a specific executable object file is loaded at depends on the memory allocated before it is loaded and when in the load order it is loaded. A statically linked executable’s address map is fixed and this is considered important in some systems. A dynamically loaded system can be loaded in a repeatable manner if the load order is the same and the initialization sequence of the system is controlled. A custom allocator may also help.
Management of dynamically loadable object files and libraries adds to the configuration management of the hosts in a project. The loadable files need to be released and tracked in a suitable configuration management process just like the base image is. Executable object files and libraries are specific to a version of RTEMS and cannot be mixed or moved and this needs to be carefully managed. Currently there are no checks an executable object file matches the version of the base image it is being loaded on. These extra configuration controlled items add to the overheads of a project and need to be considered.
Dynamically loadable systems have a number of features that benefit some systems and products. Systems can be built on a base of trusted or golden modules. A number of projects using a common base of hardware can make use of proven modules reducing the testing and qualification overhead for each new release. A tested base image with libraries for common and available boards provides a simple and fast way for new users to trial and use RTEMS.
A project can use dynamic loading during development, shipping statically linked executables in production. Hardware used by a development team can have more memory, extra media for disk drives, or a network interface.
8.7.2. Loader Interface¶
Run-time executable object file loading and management is via the standard’s based calls provided by the header file <dlfcn.h>. The details of the calls follow.
void* dlopen(const char* path, int mode);The
dlopen()function makes the symbols (function identifiers and data object identifiers) in the executable object file specified by file available to the calling program.The executable object files eligible for this operation are in the ELF format.
The link loader may load embedded dependencies in executable object files. In such cases, a
dlopen()operation may load those dependencies in addition to the executable object file specified by file.A successful
dlopen()returns a handle which the caller may use on subsequent calls todlsym(),dlinfo()anddlclose().The value of the handle should not be interpreted in any way by the caller.
Subsequent calls to
dlopen()for the same executable object file increases the references to it.The file argument is used to construct a pathname to the executable object file or archive library of executable object files. If the file argument contains a colon (
:) the name of the executable object file in the library follows and this file name may optionally end with@followed by a number which is the absolute offset in the library file where the executable object file starts. If an executable object file is not detected at the offset the archive library’s file table is searched.If file is a null pointer,
dlopen()returns a global symbol table handle. This handle provides access to the global symbols from an ordered set of executable object files consisting of the original base image file, the set of executable object files loaded usingdlopen()operations with theRTLD_GLOBALflag, and any dependencies loaded. As the latter sets of executable object files can change during execution, the set of symbols made available by this handle can also change dynamically.Only a single copy of an executable object file is brought into the address space, even if
dlopen()is invoked multiple times in reference to the executable object file, and even if different pathnames are used to reference the executable object file.Unresolved external symbols do not cause an error to be returned allowing the loading of jointly dependent executable object files.
If
dlopen()fails, it returns a null pointer, and sets an error condition which may be interrogated withdlerror().The mode parameter describes how
dlopen()operates upon file with respect to the processing of relocations and the scope of visibility of the symbols provided within file. When an executable object file is brought into the address space, it may contain references to symbols whose addresses are not known until the executable object file is loaded.If a loaded executable object file and any dependent executable object files loaded with it contain any initialiser functions, they are called in the order loaded before
dlopen()returns.The modes
RTLD_LAZYandRTLD_NOWdo not effect the type of relocation performed, it is same for both modes. All relocations of an executable object file and any dependent executable object files loaded with it are completed before thedlopen()call returns. The execution performance of the code loaded can be considered deterministic oncedlopen()has returned.Any executable object file loaded by
dlopen()can reference global symbols in the base image, any executable object files loaded included in the samedlopen()invocation, and any executable object files that were loaded in anydlopen()invocation and which specified theRTLD_GLOBALflag. To determine the scope of visibility for the symbols loaded with adlopen()invocation, the mode parameter should be a bitwise-inclusiveORwith one of the following values:RTLD_GLOBALThe executable object file’s symbols are made available for relocation processing of any other executable object file. In addition, symbol lookup using
dlopen(NULL,mode)and an associateddlsym()allows executable object files loaded with this mode to be searched.RTLD_LOCALThe executable object file’s symbols shall not be made available for relocation processing of any other executable object files.
If neither
RTLD_GLOBALnorRTLD_LOCALis specified, the default behavior is unspecified.If
RTLD_GLOBALhas been specified, the executable object file maintains it’sRTLD_GLOBALstatus regardless of any previous or future specification ofRTLD_LOCAL, as long as the executable object file remains in the address space.Symbols introduced through calls to
dlopen()may be used in relocation activities. Symbols that duplicate symbols already defined by the base image or previousdlopen()calls are treated as an error and the object file is not loaded. Symbols introduced through loading dependent executable object files are ignored or not loaded depending on the method used to build the executable object files.The symbols introduced by
dlopen()operations and available throughdlsym()are at a minimum those which are exported as identifiers of global scope by the executable object file. Typically, such identifiers shall be those that were specified in (for example) C source code as havingexternlinkage.
int dlclose(void* handle);Releases a reference to the executable object file referenced by handle. If the reference count drops to
0, the executable object file’s global symbol table is made unavailable. When all references to the global symbols the executable object file provided have been removed the object file is removed from the address space.If the executable object being removed has any termination routines in it they are called.
void* dlsym(void* handle, const char* symbol);The
dlsym()function obtains the address of a symbol (a function identifier or a data object identifier) defined in the symbol table identified by the handle argument. The handle argument is a symbol table handle returned from a call todlopen()(and which has not since been released by a call todlclose()), and name is the symbol’s name as a character string. The return value fromdlsym(), cast to a pointer to the type of the named symbol, can be used to call (in the case of a function) or access the contents of (in the case of a data object) the named symbol.The
dlsym()function searches for the named symbol in the symbol table referenced by handle and returns the address of the code or data location specified by the null-terminated character string symbol. Which libraries and objects are searched depends on the handle parameter.Upon successful completion, if name names a function identifier,
dlsym()returns the address of the function converted from type pointer to function to type pointer tovoid; otherwise,dlsym()shall return the address of the data object associated with the data object identifier named by name converted from a pointer to the type of the data object to a pointer tovoid. If handle does not refer to a valid symbol table handle or if the symbol named by name cannot be found in the symbol table associated with handle,dlsym()shall return a null pointer.
int dlinfo(void* handle, int request, void* args);
The
dlinfo()function provides information about dynamically loaded object. The action taken bydlinfo()and exact meaning and type of the argument args depend on value of the request argument provided by the caller.
RTLD_DI_UNRESOLVEDReturn
1in an indexer value pointed to by args if the symbol table handle has unresolved relocation records to symbols. If the handle is the global symbol table handle orRTLD_SELFreturn1if any unresolved relocation records to symbols are present in any loaded executable object files..
const char *dlerror(void);The
dlerror()function returns a null-terminated character string (with no trailing<newline>) that describes the last error that occurred during dynamic linking processing. If no dynamic linking errors have occurred since the last invocation ofdlerror(),dlerror()returnsNULL. Thus, invokingdlerror()a second time, immediately following a prior invocation, results inNULLbeing returned.
This example opens an object file, checks for any unresolved symbols the object file may have, locates a global symbol in the object file, calls it then closes the object file:
#include <stdbool.h>
#include <stdio.h>
#include <dlfcn.h>
typedef int (*call_sig)(void);
bool load_object (void)
{
void* handle;
call_sig call;
int unresolved;
handle = dlopen ("/code.o", RTLD_NOW | RTLD_GLOBAL);
if (handle == NULL)
{
printf ("dlopen failed: %s\n", dlerror ());
return false;
}
if (dlinfo (handle, RTLD_DI_UNRESOLVED, &unresolved) < 0)
{
printf ("dlinfo failed: %s\n", dlerror ());
dlclose (handle);
return false;
}
if (unresolved != 0)
{
printf ("object.o has unresolved external symbols\n");
dlclose (handle);
return false;
}
call = dlsym (handle, "foo");
if (call == NULL)
{
printf("dlsym failed: symbol 'foo' not found\n");
dlclose (handle);
return false;
}
printf ("'foo()' returns: %i\n", call ());
if (dlclose (handle) < 0)
{
printf("dlclose failed: %s\n", dlerror());
return false;
}
return true;
}
8.7.3. Symbols¶
The RTEMS link editor manages the symbols for the base image and all resident executable object files. A symbol is an identifier string and a pointer value to a function identifier or a data object identifier. The symbols held in the symbol tables are used in the relocation of executable object files or they can be accessed by application code using the dlsym() call.
An executable object file’s symbols are removed from the global symbol table when it is closed or orphaned. An executale object file cannot be unloaded if a symbol it provides is referenced by another object and that object is still resident. An executable object file that has no references to any of its symbols and was not explicitly loaded using the dlopen() call is orphaned and automatically removed from the address space.
8.7.3.1. Base Image Symbols¶
The base image symbol table provides access to the function and data objects statically linked into the base image. Loaded executable object files can be directly linked to the code and data resident in the base image.
A statically linked RTEMS executable does not contain a symbol table, it has to be generated and either embedded into the executable or loaded as a specially created executable object file.
The base image symbol table is dependent on the contents of the base image and this is not known until it has been linked. This means the base image symbol table needs to be constructed after the base image executable has been linked and the list of global symbols is known.
The RTEMS Tools command rtems-syms (see RTEMS Symbols) extracts the global and weak symbols from an RTEMS static executable file, creates a C file and compiles it creating a relocatable executable object file. This file can be linked with the static executable’s object files and libraries to create a static executables with an embedded symbol table or the executable file can be loaded dynamically at run-time. The following needs to be observed:
The option
-eor--embeddedto rtems-syms creates an executable object file to be embedded in the base image and not providing either of these options creates a symbols executable object file that is loaded at run-time. The same executable object file cannot be used to embedded or load.The target C compiler and machine options need to be provided to make sure the correct ABI for the target is used. See Machine Flags and ABI.
8.7.3.2. Embedded Symbols¶
An embedded symbol table is embedded within the base image executable file and loaded when the static executable is loaded into memory by the bootloader. The symbol table is automatically added to the link editor’s global symbol table when the first executable object file is loaded.
The process to embed the symbol table requires linking the base image twice. The first link is to create a static executable that collects together the symbols to make the symbol table. The RTEMS Tools command rtems-syms extracts the global and weak symbols from the static executable ELF file, creates a C file and compiles it to create an executable object file. The base image is linked a second time and this time the symbol table executable object file is added to the list of object files.
Embedding the symbol table means the chances of the symbol table and base image not matching is low, however it also means the symbol table is always present in the kernel image when dynamic loading may be optional. A project’s build system is made more complex as it needs to have extra steps to link a second time.
This example shows creating an embedded symbol table object file and linking it into the base image.
$ sparc-rtems5-gcc -mcpu=cypress foo.o -lrtemsbsp -lrtemscpu -o foo.pre
$ rtems-syms -e -C sparc-rtems5-gcc -c "-mcpu=cypress" -o foo-sym.o foo.pre
$ sparc-rtems5-gcc -mcpu=cypress foo.o foo-sym.o -lrtemsbsp -lrtemscpu -o foo.exe
The link command line steps in this example are not complete.
8.7.3.3. Loadable Symbols¶
A run-time loaded symbol table is the default for the command rtems-syms. The symbol table executable object file is packaged with the other files to be dynamically loaded at run-time and placed on the target’s file system. It needs to be loaded before any other executable object file are loaded or unresolved symbols can occur that will not be resolved.
A run-time loaded symbol table does not consume any target resources until it is loaded. This is useful in a system that optionally needs to dynamically load code, for example as a development environment. The symbol table executable needs to exactly match the base image loading it or the behavior is unpredictable. No checks are made.
The example shows creating and loading a symbol table executable object file. First create the symbol table’s executable object file:
$ sparc-rtems5-gcc -mcpu=cypress foo.o -lrtemsbsp -lrtemscpu -o foo.exe
$ rtems-syms -C sparc-rtems5-gcc -c "-mcpu=cypress" -o foo-sym.o foo.exe
The link command line steps in this example are not complete.
Load the symbol table:
#include <stdbool.h>
#include <stdio.h>
#include <dlfcn.h>
bool load (void)
{
void* handle = dlopen ("/foo-sym.o", RTLD_NOW | RTLD_GLOBAL);
if (handle == NULL)
{
printf ("failed to load the symbol table: %s\n", dlerror ());
return false;
}
return true;
}
8.7.4. Unresolved Symbols¶
The RTEMS link editor does not return an error when an executable object file is loaded with unresolved symbols. This allows dependent object files to be loaded. For example an executable object file foo.o contains the function foo() and that function calls bar() and an executable object file bar.o contains a function bar() that calls the function foo(). Either of these executable object files can be loaded first as long both are loaded before any symbols are accessed.
The link editor defers the resolution of unresolved symbols until the symbol is available in the global symbol table. Executing code or accessing data in a loaded executable object file with unresolved external symbols results in unpredictable behavior.
All unresolved symbols are checked after an executable object file has been loaded. If a symbol is found and resolved any relocations that reference the symbol are fixed. If valid library files have been configured the symbol table’s of each library are searched and if the symbol is found the dependent executable object file is loaded. This process repeats until no more symbols can be resolved.
The dlinfo() call can be used to see if a loaded executable object file has any unresolved symbols:
#include <stdbool.h>
#include <stdio.h>
#include <dlfcn.h>
bool has_unresolved(void* handle)
{
int unresolved;
if (dlinfo (handle, RTLD_DI_UNRESOLVED, &unresolved) < 0)
{
printf ("dlinfo failed: %s\n", dlerror ());
return false;
}
return unresolved != 0;
}
The handle RTLD_SELF checks for any unresolved symbols in all resident object files:
if (has_unresolved(RTLD_SELF))
printf("system has unsolved symbols\n");
8.7.5. Libraries¶
The RTEMS link editor supports loading executable object files from libraries. Executable object files can be explicitly loaded from a library using a specific path to dlopen() and treated the same as loading a stand alone executable object file. Libraries can be searched and an executable object file containing the search symbol can be loaded automatically as a dependent executable object file. A dependent executable object file loaded from a library with no symbol references to it’s symbols is orphaned and automatically unloaded and removed from the address space.
Loading Executable Object Files¶
A library is an archive format file created using the RTEMS architecture prefixed ar command. The RTEMS tool suite provides the ar program and system libraries such as libc.a and libm.a for each architecture and ABI. Libraries used by the RTEMS link editor for searching must contain a symbol table created by the ranlib program from the RTEMS tool suite.
Searching a library’s symbol table and loading an executable object file containing the symbol is called dependent loading. Dependent loading provides a simple way to manage the dependencies when loading an executable object file. If code in an executable object file references functions or data objects that are part of a library and the symbols are not part of the base image those symbols will not resolve unless the library is on the target and available for searching and loading. Dependent loading from libraries on the target provides a simple and understandable way to manage the dependency issue between the base image, loaded code and the system libraries.
The RTEMS link editor checks for the configuration file /etc/libdl.conf on each call to dlopen(). If the file has changed since the last check it is loaded again and the contents processed. The file format is:
Comments start with the
#character.A line is a wildcard path of libraries to search for. The wildcard search uses the
fnmatch()call. Thefnmatch()function matches patterns according to the rules used by a shell.
Files that match the search pattern are verified as a library and if a symbol table is found it is loaded and the symbols it contains made search-able.
A call to dlopen() checks the global symbols table and any references to relocation symbols not found are unresolved and added to the unresolved symbol table. Once the executable object file is loaded the link editor attempts to resolve any unresolved symbols. The unresolved symbol resolver checks each unresolved symbol against the global symbol table and if not found the available library symbol tables are searched. If a symbol is found in a library the dependent executable object file is loaded. The process repeats until all unresolved symbols have been resolved and the remaining unresolved symbols are not in the global symbol table or any libraries. The loading of a library executable object file will resolve at least one symbol and it may add more unresolved symbols requiring further searching of the libraries.
A library of executable object files built by the RTEMS Tool suite can contain debug information and this should be stripped before loading on to the target. The tool suite’s command strip can strip all the object files in a library with a single command.
$ sparc-rtems5-strip libc.a
8.7.6. Large Memory¶
The RTEMS link editor supports large memory relocations. Some architectures have instructions where the relative branch or jump offset from the instruction to the target address is limited. These instructions provide improved performance because less code generated compared to larger instructions which contain full address space references. The compact code helps lower cache pressure as well and providing improved performance for localalized functions and loops. The compiler defaults to generating the smaller instruction and as the final address map not known when generating the code, linkers need to provide glue code to extend the small address range to the enitre address space. This is called a trampoline. A trampoline is transparent to the execution of the code.
The link editor parses an executable object file’s relocation records to determine the number of trampolines needed. Added to this value are all unresolved symbols present in an executable object file after it is loaded. There is a slot allocated even if the symbol ends up being within range as there is no way to determine a symbol’s address until it is loaded and the range calculated.
The trampoline table is allocated a separate block of memory to the executable object file’s text, data and constants memory. The trampoline parsing requires the executable object file’s instructions (text) be in memory as the instructions are inspected by the architecture specific relocation support to determine an instruction’s range. As a result the allocation for the trampoline table has to occur after the text memory has been allocated. Most instructions have relative offsets and the trampoline table is allocated at one end limiting the size of an object to half the maximum range.
Trampolines support is available for the ARM and PowerPC architectures. The SPARC and Intel x86 architectures do not need trampolines and MIPS needs support added.
8.7.7. Allocator¶
The RTEMS link editor supports custom allocators. A custom allocator lets you manage the memory used by the RTEMS link editor as it runs. Allocators could provide:
Support for the various types of memory that can be allocated allowing specialised target support for specific use cases.
Locking of read-only memory. The link editor unlocks read-only memory when it needs to write to it.
Separation of memory holding code and data from the heap.
The allocator can be hooked using the rtems_rtl_alloc_hook call before any calls to dlopen() are made. The hook call returns the current allocate allowing the allocators to be chained.
The default allocator uses the heap.
The allocator tags specify the type of memory the allocator is handling. The tag used to allocate memory at an address must be used when making allocator calls. The rtems_rtl_alloc_tags are:
RTEMS_RTL_ALLOC_OBJECTAllocate a generic object. The link editor uses this memory for data structures it uses to manage the linking process and resident executable object files.
RTEMS_RTL_ALLOC_SYMBOLAllocate memory to hold symbol data.
RTEMS_RTL_ALLOC_EXTERNALAllocate memory for unresolved external symbols.
RTEMS_RTL_ALLOC_READAllocate memory for read-only data such as constants and exception tables.
RTEMS_RTL_ALLOC_READ_WRITEAllocate memory for read-write data such as a initialised, uninitialized and common variables.
RTEMS_RTL_ALLOC_READ_EXECAllocate memory for code to be executed in. The address space is configure for read and execute.
The commands are used to control the action the allocator performs. The rtems_rtl_alloc_cmd are:
RTEMS_RTL_ALLOC_NEWAllocate memory of the
tagtype. ReturnsNULLif the allocation fails.
RTEMS_RTL_ALLOC_DELDelete a previous allocation freeing the memory. The
taghas to match address of the memory being deleted.
RTEMS_RTL_ALLOC_WR_ENABLEEnable writes to a region of memory previously allocated. The
taghas to match the address of the memory being write enabled. The link editor may call issue this command for memory that is already write enabled.
RTEMS_RTL_ALLOC_WR_DISABLEDisable writes to a region of memory previously allocated. The
taghas to match address of the memory being write disabled. The link editor may call issue this command for memory that is writable and not to be write disabled. The allocator need to manage this case.
The allocator handler is a single call to handle all allocator requests. The handler called on evey allocation action made by the link editor. The type of the function you need is:
typedef void (*rtems_rtl_allocator)(rtems_rtl_alloc_cmd cmd,
rtems_rtl_alloc_tag tag,
void** address,
size_t size);
The arguments are:
cmdThe command to action. See rtems_rtl_alloc_cmd.
tag
The type of memory the command is for. The
tagmust match the address for commands other thanRTEMS_RTL_ALLOC_OBJECT. See rtems_rtl_alloc_tags.
addressPointer to the address. This is set of the
RTEMS_RTL_ALLOC_OBJECTcommand and read for the other commands. Thetagmust match the address for commands that read the address from the pointer.sizeThe size of the memory to allocate. This is only valid for the
RTEMS_RTL_ALLOC_OBJECTcommand.
The call to hook the allocator is:
rtems_rtl_allocator rtems_rtl_alloc_hook (rtems_rtl_allocator handler);
The current allocator is returned. You can provide a full allocator or you can filter commands.
8.7.8. Languages¶
C is supported.
C++ is supported. Initializer functions are called when an object is loaded and finalizer functions are called before it is unloaded and removed. Static constructions are initializer functions and static destructors are finalizer functions.
C++ exceptions are handled across modules. The compiler generated exception tables present in an executable object file are registered with the architecture specific mechanism when loaded and deregistered when unloaded. An exception thrown in loaded code can be caught in the base image or another loaded module. If you are using C++ and exceptions it is recommended some exception code is added to the base image to place the architecture specific support in the base image.
8.7.9. Thread Local Storage¶
Thread local storage (TLS) is currenly not supported by the RTEMS link editor. The RTEMS executive needs to have a special allocator added to manage dynamically allocating TLS variables in a thread.
If you need TLS support in dynamically loaded code please consider the RTEMS support options.
8.7.10. Architectures¶
The following architectures are supported:
ARM
Blackfin
H8300
Intel x86 (i386)
LM32
M68K
MIPS
Moxie
PowerPC
SPARC
V850
8.7.10.1. ARM¶
The ARM relocation backend supports veneers which is trampolines.
The veneer implementation is a single instruction and a 32bit target address making the overhead 8 bytes for each veneer. The performance overhead is a single instruction.
8.7.10.2. PowerPC¶
The PowerPC relocation backend support trampolines and small data.
The trampoline is four instructions and uses register 12 which the PowerPC ABI reserves for scratch use. The implementation loads the counter register and branches to the address it contains. The trampoline size is 16 bytes. The performance overhead is four instructions.
The PowerPC relocation backend also supports small data. The sections of an executable object file are parsed and small data are tagged as needing architecture specific allocations. These sections are not allocated as part of the standard section allocation. Small data sections are allocated in the global small data region of memory. The size of this region is defined in the BSP’s linker command file by setting bsp_section_small_data_area_size variable:
bsp_section_small_data_area_size = 65536;
The maximum size of the small data region is 65536 bytes. It is recommended code built for loading uses the same settings for small base as the base image.
8.8. Device Tree¶
A Device Tree is a data structure that is used to describe properties of non-discoverable hardware instead of hardcoding them in the kernel. The device tree data is generally stored in a .dts or a Device Tree Source (DTS) file. This file is then compiled into a binary format called Device Tree Blob (DTB) with .dtb extension. RTEMS preferably uses a DTB built from the FreeBSD source tree matching the freebsd-org HEAD commit hash in libBSD.
8.8.1. Building the DTB¶
A single DTB file can be built using the dtc tool in libfdt using the following command:
dtc -@ -I dts -O dtb -o my-devicetree.dtb my-devicetree.dts
For building the DTB from the FreeBSD source, the make_dtb.sh script from freebsd/sys/tools/fdt must be used as most of the DTS files in FreeBSD have included .dtsi files from their source tree. An example is given below as a reference for how to build the device tree from the FreeBSD source.
NOTE: The following example uses FreeBSD master branch from github mirror as an example. It is advised to always use the source from the commit matching the freebsd-org HEAD in libBSD.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | #We're using the script from freebsd/sys/tools/make_dtb.sh
#Target device: Beaglebone Black.
#Architecture: Arm.
#DTS source name: am335x-boneblack.dts
#The make_dtb.sh script uses environment variable MACHINE
export MACHINE='arm'
SCRIPT_DIR=$HOME/freebsd/sys/tools/fdt
#The arguments to the script are
# $1 -> Build Tree (This is the path to freebsd/sys/ directory)
# $2 -> DTS source file
# $3 -> output path of the DTB file
${SCRIPT_DIR}/make_dtb.sh ${SCRIPT_DIR}/../../ \
${SCRIPT_DIR}/../../gnu/dts/arm/am335x-boneblack.dts \
$(pwd)
|
8.8.2. Using Device Tree Overlay¶
Device tree overlay is used either to add properties or devices to the existing device tree. Adding any property to DTS using an overlay will override the current values in the DTB. The Overlays enable us to modify the device tree using a small maintainable plugin without having to edit the whole Base Tree.
There are two ways of applying an overlay on top of the built DTB.
Use fdtoverlay from libfdt
Add the overlay in the root partition of the SD card and apply it using U-Boot
The fdtoverlay command can be used as follows:
fdtoverlay -i my-base-tree.dtb -o output-tree.dtb my-overlay.dtbo
To apply it from U-Boot during system initialization we have to add the device tree overlay file in the root directory of the SD card and use U-Boot commands to apply the overlay.
Below is given the series of U-Boot commands that can be used to apply the overlay, given that the overlay blob (.dtbo) file is already in the card.
fatload mmc 0:1 0x80800000 rtems-app.img
fatload mmc 0:1 0x88000000 my-base-tree.dtb
fdt addr 0x88000000
fatload mmc 0:1 0x88100000 my-overlay.dtbo
fdt resize 0x1000
fdt apply 0x88100000
bootm 0x80800000-0x88000000
9. Testing¶
RTEMS developers run test executables when adding new features or testing a bug fix. All tests are run to make sure changes do not introduce regressions. Users can run the RTEMS tests to be certain the build of the kernel they have is functioning.
The section describes using and configuring the RTEMS Tester and RTEMS Run tools, the types of laboratory set ups supported and how to add your BSP to the framework. The tools command line interfaces are detailed in RTEMS Tester and Run.
An RTEMS Test is an RTEMS executable where the application code is a test. Tests in RTEMS print banners to the console to indicate the configuration of the test and if it has start and finished.
The RTEMS Tools Project provides the RTEMS Tester and RTEMS Run tools. The RTEMS Tester command is rtems-test and the RTEMS Run command is rtems-run. These commands manage the complexity of running embedded executables. The commands provide a consistent command line interface to a testing framework that supports the various run time and testing scenarios we encounter such as simulators, GDB and executing directly on target hardware.
The RTEMS kernel code contains an extensive set of tests to exercise and test the RTEMS kernel. The tests check functionality, provide coverage testing and make sure the kernel is operating as intended on your target system. The testsuite has support to make creating a test simple and uniform.
The tests are built by adding --enable-tests to the RTEMS build configuration command line. There are over 600 tests and building them does extend the RTEMS kernel’s build time and use more disk space but it worth building and running them. The RTEMS test executables have the .exe file extension.
9.1. Test Banners¶
All test output banners or strings are embedded in each test and the test outputs the banners to the BSP’s console as it executes. The RTEMS Tester captures the BSP’s console and uses this information to manage the state of the executing test. The banner strings are:
*** BEGIN TEST <name> ***The test has loaded, RTEMS has initialized and the test specific code is about to start executing. The
<name>field is the name of the test. The test name is internal to the test and may not match the name of the executable. The test name is informative and not used by the RTEMS Tester.
*** END TEST <name> ***The test has finished without error and has passed. The
<name>field is the name of the test. See the Test Begin Banner for details about the name.
*** TEST VERSION: <version>The test prints the RTEMS version return by the RTEMS Version API as
<version>. All tests must match the first test’s version or the Wrong Version error count is incremented.
*** TEST STATE: <state>The test is tagged in the RTEMS sources with a special
<state>for this BSP. See Test States for the list of possible states. The state banner lets the RTEMS Tester categorize and manage the test. For example a user input test typically needing user interaction may never complete producing an invalid test result. A user input test is terminated to avoid extended delays in a long test run.
*** TEST BUILD: <build>The test prints the RTEMS build as a space separated series of labels as
<build>. The build labels are created from the configuration settings in the Super Score header filertems/score/cputops.h. All tests must match the first test’s build or the Wrong Build error count is incremented.
*** TEST TOOLS: <version>The test prints the RTEMS tools version returned the GGC internal macro
_VERSION_as<version>. All tests must match the first test’s tools version string or the Wrong Tools error count is incremented.
9.2. Test Controls¶
The tests in the RTEMS kernel testsuite can be configured for each BSP. The expected state of the test can be set as well as any configuration parameters a test needs to run on a BSP.
The test states are:
passedThe test start and end banners have been sent to the console.
failureThe test start banner has been sent to the console and no end banner has been seen when a target restart is detected.
excepted-failThe test is tagged as
expected-failin the RTEMS sources for this BSP and outputs the banner*** TEST STATE: EXPECTED_FAIL. The test is known not to pass on this BSP. The RTEMS Tester will let the test run as far as it can and if the test passes it is recorded as a pass in the test results otherwise it is recorded as expected-fail.
indeterminateThe test is tagged as
indeterminatein the RTEMS sources for this BSP and outputs the banner*** TEST STATE: INDETERMINATE. The test may or may not pass so the result is not able to be determined. The RTEMS Tester will let the test run as far as it can and record the result as indeterminate.
user-inputThe test is tagged as
user-inputin the RTEMS sources and outputs the banner*** TEST STATE: USER_INPUT. The RTEMS Tester will reset the target if the target’s configuration provides a target reset command.
benchmarkThe test is tagged as
benchmarkin the RTEMS sources and outputs the banner*** TEST STATE: BENCHMARK. Benchmarks can take a while to run and performance is not regression tested in RTEMS. The RTEMS Tester will reset the target if the target’s configuration provides a target reset command.
timeoutThe test start banner has been sent to the console and no end banner is seen within the timeout period and the target has not restart. A default timeout can be set in a target configuration, a user configuration or provide on the RTEMS Tester’s command line using the
--timeoutoption.
invalidThe test did not output a start banner and the RTEMS Tester has detected the target has restarted. This means the executable did not load correctly, the RTEMS kernel did not initialize or the RTEMS kernel configuration failed for this BSP.
9.2.1. Expected Test States¶
A test’s expected state is set in the RTEMS kernel’s testsuite. The default for a tested is to pass. If a test is known to fail it can have it’s state set to expected-fail. Setting tests that are known to fail to expected-fail lets everyone know a failure is not to be countered and consider a regression.
Expected test states are listed in test configuration files
9.2.2. Test Configuration¶
Tests can be configured for each BSP using test configuration files. These files have the file extension .tcfg. The testsuite supports global test configurations in the testsuite/testdata directory. Global test states are applied to all BSPs. BSPs can provide a test configuration that applies to just that BSP and these files can include subsets of test configurations.
The configuration supports:
Including test configuration files to allow sharing of common configurations.
Excluding tests from being built that do not build for a BSP.
Setting the test state if it is not
passed.Specifing a BSP specific build configuration for a test.
The test configuration file format is:
state: test
where the state is state of the test and test is a comma separated list of tests the state applies too. The configuration format is:
flags: test: configuration
where flags is the type of flags being set, the test is a comma separate list of regular expresions that match the test the configuration is applied too and the configuration is the string of flags.
The state is one of:
includeThe test list is the name of a test configuration file to include
excludeThe tests listed are not build. This can happen if a BSP cannot support a test. For example it does not have enough memory.
expected-failThe tests listed are set to expected fail. The test will fail on the BSP being built.
user-inputThe tests listed require user input to run and are not supported by automatic testers.
indeterminateThe tests listed may pass or may not, the result is not reliable.
benchmarkThe tests listed are benchmarks. Benchmarks are flagged and not left to run to completion because they may take too long.
Specialized filtering using regular expressions is supported using:
rexcludeThe test matching the regular expression are excluded.
rincludeThe tests matching the regular expression are included.
By default all tests are included, specific excluded tests using the exclude state are excluded and cannot be included again. If a test matching a rexclude regular it is excluded unless it is included using a rinclude regular expression. For example to build only the hello sample application you can:
rexclude .*
rinclude hello
Test configuration flags can be applied to a range of tests using flags. Currently only cflags is support. Tests need to support the configuration for it to work. For example to configure a test:
cflags: tm.*: -DTEST_CONFIG=42
cflags: sp0[456]: -DA_SET_OF_TESTS=1
Flags setting are joined together and passed to the compiler’s command line. For example:
cflags: tm.*: -DTEST_CONFIG=42
cflags: tm03: -DTEST_TM03_CONFIG=1
would result in the command line to the test tm03 being:
-DTEST_CONFIG=42 -DTEST_TM03_CONFIG=1
Specific flags can be set for one test in a group. For example to set a configuration for all timer tests and a special configuraiton for one test:
cflags: (?!tm02)tm.*: -DTEST_CONFIG=one
cflags: tm02: -DTEST_CONFIG=two
9.3. Test Builds¶
The test reports the build of RTEMS being tested. The build are:
defaultThe build is the default. No RTEMS configure options have been used.
posixThe build includes the POSIX API. The RTEMS configure option
--enable-posixhas been used. Thecpuopts.hdefineRTEMS_POSIXhas defined and it true.
smpThe build is an SMP kernel. The RTEMS configure option
--enable-smphas been used. Thecpuopts.hdefineRTEMS_SMPhas defined and it true.
mpThe build is an MP kernel. The RTEMS configure option
--enable-multiprocessinghas been used. Thecpuopts.hdefineRTEMS_MULTIPROCESSINGhas defined and it true.
paravirtThe build is a paravirtualization kernel. The
cpuopts.hdefineRTEMS_PARAVIRThas defined and it true.
debugThe build includes kernel debugging support. The RTEMS configure option
--enable-debughas been used. Thecpuopts.hdefineRTEMS_DEBUGhas defined and it true.
profilingThe build include profiling support. The RTEMS configure option
--enable-profilinghas been used. Thecpuopts.hdefineRTEMS_PROFILINGhas defined and it true.
9.4. Tester Configuration¶
The RTEMS Tester and RTEMS Run are controlled by configuration data and scripts. The user specifies a BSP on the command line using the --rtems-bsp option as well as optionally specifying a user configuration file using --user-config.
The Figure RTEMS Tester and Run Configuration Files shows the various sources of configuration data and their format. The ini files are the standard INI format, the mc are the internal RTEMS Toolkit’s Macro format, and cfg is the RTEMS Toolkit’s Configuration script format, the same format used by the RTEMS Source Builder.
RTEMS Tester and Run Configuration Files¶
Configuration data is held in a macro database keyed on the macro name. Macros can be expanded in configuration scripts using the syntax %{name}. The macro database is layered using maps. The defaults and values created when a configure script runs live the in the global map. Values read from the BSP and User INI configuration files are loaded into maps based on the BSP name. This lets a single User configuration file contain specialized configuration values for a number of BSPs and the tester and run commands select the values based on the selected BSP. Macros are expanded using the BSP map first giving those values the highest priority. User defined values are loaded after the BSP configuration values overwriting them letting a user speckles a BSP’s default configuration for their local needs.
Figure RTEMS Tester and Run Configuration Load and Execute Sequence shows the configuration loading and script execution order.
RTEMS Tester and Run Configuration Load and Execute Sequence¶
9.4.1. Defaults¶
The RTEMS Tester and RTEMS Run are primed using defaults from the file rtems/testing/testing.mc. All default settings can be overridden in a BSP or User configuration file.
9.4.2. BSP and User Configuration¶
The BSP and User configuration files are INI format files. The BSP configuration file has to have an INI section that is the name of the BSP passed on the command line. The section has the following mandatory values:
bspThe name of the BSP. The BSP name is used to create a macro map to hold the BSP’s configuration data. Typically this is the same as the BSP name used on the command line.
archThe name of the BSP architecture. This is need for the GDB configuration scripts where the architecture specific GDB needs to run. It is mandatory so the arch/bsp standard RTEMS BSP string can be used.
testerThe tester or run configuration script. This is the name of the configuration script the RTEMS Tester or RTEMS Run executes as a back end. The
testervalue is typically of the form%{_rtscripts}/<script>where<script>is name of the back end script to be run.
Target commands support expansion of specific tags to provide a convenient way for users to customize a local test environment. The parameters expanded are:
@ARCH@The BSP architecture.
@BSP@The BSP’s name set by the
bspvalue.
@EXE@The executable name as an absolute path
@FEXE@The filtered executable if a
target_exe_filteris provided else the executable’s file name.
The following are optional and depend on the back end being used and the local target hardware set up:
jobsThe jobs value sets the number of jobs that can be run at once. This setting only effects the RTEMS Tester. The tester can run up to the
jobsvalue of tests concurrently. If the tester back end is a simulator running a job on each available core lowers the total test time. Overloading a machine with too many simulators running in parallel can slow down each simulation and test timeouts may be recorded.
bsp_tty_devThe BSP’s tty device. This can be a real device on the host machine the executable is being run from or it can be a telnet server and port defined using the stand host format. See Consoles for details.
target_pretest_commandThe pre-test command is a host shell command that is called before each test runs. It can be used to construct a suitable environment or image needed by a simulator or target. The RTEMS executate being run is provided as an argument and the bootloader specific format is the output.
target_posttest_commandThe post-test command is a host shell command that is called after each test has finished. It can be used to destroy any environment or image created by the pre-test command.
target_exe_filterThe target executable filter transforms the executable name into a filtered executable name. This filter lets the tester or run command track the name of any generated file a pre-test command may generate. The syntax is a simplified
sedregular expression. The first character is a delimiter and there must be 2 sections therefore 3 delimiter. The first section is a Python regular expression and the second section is plain text that replaces anywhere the regular expression matches. For example/\.exe/.exe.img/will search for.exein the executable name and replace it with.exe.img. Note, there is no need to escape the text in the second part, it is just plain test.
test_restartsThe number of restarts before the test is considered
invalid. Currently not used.
target_reset_regexThe target reset regular expression. This is a Python regular expression used to filter the console input. If a match is made something has happened during the boot process that requires a reset. The
target_reset_commandis issued to perform the reset. Typically this field looks for boot loader error messages that indicate the boot process as failed.
target_start_regex
The target start regular expression. This is a Python regular expression to filter the console input to asynchronously detect if a target has reset. If a board crashes running a test or at any point reset this filter detects the restart and ends the test with a suitable result.
target_on_commandThe target on command is a host shell command that is called before the first test. This command powers on a target. Targets should be left powered off when not running tests or the target may request TFTP downloads that are for another target interfering with those test results. We recommend you implement this command as a target off command, a pause, then a target on command.
target_off_commandThe target off command is a host shell command that is called after the last test powering off the target.
target_reset_commandThe target reset command is a host shell command that is called when the target needs to be reset. This command can power cycle the target or toggle a reset signal connected to the target. If you are power cycling a target make sure you have a suitable pause to let the target completely power down.
9.4.3. Configuration Scripts¶
Configuration scripts are provided for each supported RTEMS Tester and RTEMS Run back end and console management. The scripts are in the standard RTEMS Toolkit Configuration Script format. Please refer to the RTEMS Source Builder documentation for the basic scripting syntax and usage.
The RTEMS Tester and RTEMS Run specializes the standard configuration syntax providing a directive for the console and each supported back end. The supported directives are:
%console%execute%gdb%tftp
9.4.3.1. Console¶
The %console configures the console used to access the target’s console. The console can be a process’s stdout, a termios tty on Unix and MacOS and Telnet on all hosts. The directive accepts:
stdioThe standard output stream from the executing processing.
tty <dev> <settings>The name of the
ttyto open and use. Thettydevice or<dev>can be a termio device and the<settings>are standard termios values.The Python termios document provides details of the settings that can be controlled. The settings are a single string where prefix the value with
~negates the setting. Setting are:B115200(an example buadrate)BRKINTIGNBRKIGNCRICANONISIGIEXTENECHOCLOCALCRTSCTSVMIN=<value>VTIME=<value
A example in a configuration script is:
%define bsp_tty_dev /dev/ttyUSB2
%define bsp_tty_settings B115200,~BRKINT,IGNBRK,IGNCR,~ICANON,~ISIG,~IEXTEN,~ECHO,CLOCAL,~CRTSCTS,VMIN=1,VTIME=2
A example BSP or User configuration file is:
[bsp-special]
bsp = example-bsp
bsp_tty_dev = /dev/ttyUSB2
bsp_tty_settings = B115200,~BRKINT,IGNBRK,IGNCR,~ICANON,~ISIG,~IEXTEN,~ECHO,CLOCAL,~CRTSCTS,VMIN=1,VTIME=2
The console directive is managed in the %{_rtscripts}/console.cfg configuration script. If the %{console_stdio} is defined the console will be stdio else the console will be the BSP console or %{bsp_tty_dev}.
Telnet can be combined with the ser2net daemon to remotely access a target’s physical serial UART interface.
9.4.3.2. Execute¶
The %execute directive executes a command for each rest. The execute forks the command and arguments supplied to the execute directive and captures the stdout stream as the console. If the console directive is set to stdout the sub-processes stdout stream is used as the console.
The RTEMS Tester will run parallel tests as jobs.
An example is:
%execute %{run_cmd} %{run_opts} %{test_executable} %{test_executable_opts}
9.4.3.3. GDB¶
The %gdb directive executes GDB in the machine interface mode give the RTEMS Tester and RTEMS Run commands control. The console is taken from GDB if it is stdout.
The RTEMS Tester will run parallel tests as jobs.
An example is:
%gdb %{gdb_cmd} %{test_executable} %{gdb_script}
9.4.3.4. TFTP¶
The %tftp directive starts a TFTP session on a specified port sending the test executable to the target over a networking using the TFTP protocol.
The RTEMS Tester will run only one test at a time. There is just one physical board running the test.
An example is:
%tftp %{test_executable} %{tftp_port}
9.5. Coverage Analysis¶
RTEMS is used in many critical systems. It is important that the RTEMS Project ensure that the RTEMS product is tested as thoroughly as possible. With this goal in mind, the RTEMS test suite was expanded with the goal that 100% of the RTEMS executive is tested.
RTEMS-TESTER takes the following arguments to produce coverage reports:
- –coverage :
When the coverage option is enabled the tester produces coverage reports for all the symbols in cpukit. To generate a coverage report for a specific symbol-set ( e.g.: score) the symbol-set is passed as an argument to the option, e.g.: –coverage=score.
- –no-clean :
Tells the script not to delete the .cov trace files generated while running the coverage. These trace files are used for debugging purposes and will not be needed for a normal user.
For example: To generate a coverage report of hello.exe for leon3 on SIS, the following command is used:
rtems-test \
--rtems-tools=$HOME/development/rtems/5 \
--log=coverage_analysis.log \
--no-clean \
--coverage \
--rtems-bsp=leon3-sis-cov \
$HOME/development/rtems/kernel/leon3/sparc-rtems5/c/leon3/testsuites/samples/hello.exe
The command will create the coverage report in the following tree structure:
├── coverage_analysis.log
├── leon3-sis-coverage
│ └── score
│ ├── annotated.html
│ ├── annotated.txt
│ ├── branch.html
│ ├── branch.txt
│ ├── covoar.css
│ ├── ExplanationsNotFound.txt
│ ├──
│ ├── no_range_uncovered.html
│ ├── no_range_uncovered.txt
│ ├── NotReferenced.html
│ ├── sizes.html
│ ├── sizes.txt
│ ├── summary.txt
│ ├── symbolSummary.html
│ ├── symbolSummary.txt
│ ├── table.js
│ ├── uncovered.html
│ └── uncovered.txt
└── leon3-sis-report.html
The html on top of the directory, i.e., leon3-sis-report.html is the top level navigation for the coverage analysis report and will let the user browse through all the generated reports from different subsystems.
9.6. Consoles¶
The RTEMS Tester uses the target’s console output to determine the state of a test. Console interfaces vary depending on the testing mode, the BSP, and the target hardware.
Consoles for simulator work best if mapped to the simulator’s stdout interface. The RTEMS Tester can capture and process the stdout data from a simulator while it is running.
Target hardware console interfaces can vary. The most universal and stable interface target hardware is a UART interface. There are a number of physical interfaces for UART data these days. They are:
RS232
TTL
USB
RS232 is still present on a number of targets. The best solution is to use a RS232 to USB pod and convert the port to USB.
TTL is common on a number of boards where cost is important. A console interface is typically a development tool and removing the extra devices need to convert the signal to RS232 or directly to USB is not needed on production builds of the target. There is a standard header pin out for TTL UART consoles and you can purchase low cost cables with the header and a built in UART to USB converter. The cables come is different voltage levels so make sure you check and use the correct voltage level.
The USB interface on a target is typcially a slave or OTG interface and all you need to a standard USB cable.
We recommend a low cost and low power device to be a terminal server. A Raspberry Pi or similar low cost computer running Linux can be set up quickly and with a powered USB hub and can support a number of USB UART ports. A USB hub with a high power port is recommended that can suppy the Raspberry Pi.
The open source daemon ser2net is easy to configure to map the USB UART ports to the Telnet protocol. There is no need for security because a typical test environment is part of a lab network that should be partitioned off from an enginnering or corportate network and not directly connected to the internet.
A test set up like this lets you place a terminal server close to your target hardware providing you with the flexibility to select where you run the RTEMS Tester. It could be your desktop or an expensive fast host machine in a server rack. None of this equipment needs to directly interface to the target hardware.
The RTEMS Tester directly supports the telnet protcol as a console and can interface to the ser1net server. The telnet console will poll the server waiting for the remote port to connect. If the terminal server ser2net does not have a tty device it will not listen on the port assigned to that tty. A USB tty can come and go depending on the power state of the hardware and the target hardware’s design and this can cause timing issues if the target hardware is power cycled as part of a reset process.
9.7. Simulation¶
Simulation is a important regression and development tool for RTEMS. Developers use simulation to work on core parts of RTEMS as it provides excellent debugging supporting. Simulation run via the RTEMS Tester allows a test to run on each core of your testing host machine lower the time to run all tests.
RTEMS Tester Simulation¶
The RTEMS Tester Simulation figure shows the structure of RTEMS Testing using simulation. The executables are built and the rtems-test command is run from the top of the build directory. The RTEMS Tester executes the BSP specific simulator for each test capturing the output
9.8. GDB and JTAG¶
GDB with JTAG provides a low level way to runs tests on hardware with limited resources. The RTEMS Tester runs and controls an instance of GDB per test and GDB connects via the GDB remote protocol to a GDB server that interfaces to the JTAG port of a target.
RTEMS Tester using GDB and JTAG¶
The Figure RTEMS Tester using GDB and JTAG shows the structure of RTEMS Testing using GDB and JTAG. The executables are built and the rtems-test command is run from the top of the build directory. The RTEMS Tester executes the BSP architecture’s GDB and expects the user to provide a gdb-script to connect t the JTAG GDB server.
9.9. TFTP and U-Boot¶
TFTP and U-Boot provides a simple way to test RTEMS on a network capable target. The RTEMS Tester starts a TFTP server for each test and the target’s boot monitor, in this case U-Boot request a file, any file, which the TFTP server supplies. U-Boot loads the executable and boots it using a standard U-Boot script.
RTEMS Tester using TFTP and U-Boot.¶
The Figure RTEMS Tester using TFTP and U-Boot. figure shows the structure and control flow of the RTEMS Tester using TFTP and U-boot. The executables are built and the rtems-test command is run from the top of the build directory.
This test mode can only support a single test job running at once. You cannot add more test target hardware and run the tests in parallel.
9.9.1. Target Hardware¶
The RTEMS Tester TFTP and U-Boot method of testing requires:
A target with network interface.
U-Boot, iPXE or similar boot loader with network driver support for your target hardware and support for the TFTP protocol.
Network power of IO switch.
Network DHCP server.
Console interface cable that matches your target’s console UART interface.
Telnet terminal server. See Consoles.
The network power or IO switch is a device that can control power or an IO pin over a network connection using a script-able protocol such as Telnet or curl. This device can be used with the target control commands.
9.9.1.1. U-Boot Set Up¶
Obtain a working image of the U-Boot boot loader for your target. We suggest you follow the instructions for you target.
Configure U-Boot to network boot using the TFTP protocol. This is U-Boot script for a Zedboard:
loadaddr=0x02000000
uenvcmd=echo Booting RTEMS Zed from net; set autoload no; dhcp; set serverip 10.10.5.2; tftpboot zed/rtems.img; bootm; reset;
The load address variable loadaddr is specific to the Zedboard and can be found in the various examples scripts on the internet. The script then sets U-Boot environment variable autoload to no causing DHCP to only request a DHCP lease from the DHCP server. The script sets the serverip to the host that will be running the RTEMS Tester then issues a TFTP request. The file name can be anything because the RTEMS Tester ignores it sending the executable image under test. Finally the script boots the download executable and if that fails the catch all reset resets the board and starts the boot process over.
Test the target boots and U-Boot runs and obtains a valid DHCP lease. Manually connect the console’s telnet port.
9.9.2. BSP Configuration¶
The BSP’s configuration file must contain the standard fields:
bsparchjobs- Must be set to1.tester- Set to%{_rtscripts}/tftp.cfg
For example the Zedboard’s configuration is:
[xilinx_zynq_zedboard]
bsp = xilinx_zynq_zedboard
arch = arm
jobs = 1
tester = %{_rtscripts}/tftp.cfg
The TFTP configuration supports the following field’s:
bsp_tty_devThe target’s tty console. For telnet this is a host and port pair written in the standard networking format, for example
serserver:12345.test_restartsThe number of restarts before the test is considered
invalid.target_reset_regexThe target reset regular expression. This is a Python regular expression used to filter the console input. If a match is made something has happened during the boot process that requires a reset. The
target_reset_commandis issued to perform the reset. This field is typically looks for boot loader error messages that indicate the boot process as failed.target_start_regexThe target start regular expression. This also a Python regular expression to filter the console input to detect if a target has reset. If a board crashes running a test or at any point in time and reset this filter detects this as happened and end the test with a suitable result.
target_on_commandThe target on command is a host shell command that is called before the first test. This command powers on a target. Targets should be left powered off when not running tests or the target may request TFTP downloads that are for another target interfering with those test results. We recommend you implement this command as a target off command, a pause, then a target on command.
target_off_commandThe target off command is a host shell command that is called after the last test powering off the target.
target_reset_commandThe target reset command is a host shell command that is called when the target needs to be reset. This command can power cycle the target or toggle a reset signal connected to the target. If you are power cycling a target make sure you have a suitable pause to let the target completely power down.
target_pretest_commandThe target pretest command is a host shell comment that is called before the test is run
The commands in the listed fields can include parameters that are substituted. The parameters are:
@ARCH@The BSP architecture
@BSP@The BSP’s name
@EXE@The executable name.
@FEXE@The
- . The
@ARCHis the
substituted
Some of these field are normally provided by a user’s configuration. To do this use:
requires = bsp_tty_dev, target_on_command, target_off_command, target_reset_command
The requires value requires the user provide these settings in their configuration file.
The Zedboard’s configuration file is:
[xilinx_zynq_zedboard]
bsp = xilinx_zynq_zedboard
arch = arm
jobs = 1
tester = %{_rtscripts}/tftp.cfg
test_restarts = 3
target_reset_regex = ^No ethernet found.*|^BOOTP broadcast 6.*|^.+complete\.+ TIMEOUT.*
target_start_regex = ^U-Boot SPL .*
requires = target_on_command, target_off_command, target_reset_command, bsp_tty_dev
The target_start_regex searches for U-Boot’s first console message. This indicate the board can restarted.
The target_reset_regex checks if no ethernet interface is found. This can happen if U-Boot cannot detect the PHY device. It also checks if too many DHCP requests happen and finally a check is made for any timeouts reported by U-Boot.
An example of a user configuration for the Zedboard is:
[xilinx_zynq_zedboard]
bsp_tty_dev = selserver:12345
target_pretest_command = zynq-mkimg @EXE@
target_exe_filter = /\.exe/.exe.img/
target_on_command = power-ctl toggle-on 1 4
target_off_command = power-ctl off 1
target_reset_command = power-ctl toggle-on 1 3
9.9.3. TFTP Sequences¶
Running a large number of tests on real hardware exposes a range of issues and RTEMS Tester is designed to be tolerant of failures in booting or loading that can happen, for example a hardware design. These sequence diagrams document some of the sequences that can occur when errors happen.
The simplest sequence is running a test. The target is powered on, the test is loaded and executed and a pass or fail is determined:
Test Pass and Fail Sequences¶
The target start filter triggers if a start condition is detected. This can happen if the board crashes or resets with no output. If this happens repeatedly the test result is invalid:
Target Start Filter Trigger¶
The reset filter triggers if an error condition is found such as the bootloader not being able to load the test executable. If the filter triggers the target_reset_command is run:
Target Reset Filter Trigger¶
If the RTEMS Tester does not detect a test has started it can restart the test by resetting the target. The reset command can toggle an IO pin connected to reset, request a JTAG pod issue a reset or turn the power off and on:
Target Timeout¶
10. Tracing¶
RTEMS Tracing Framework is an on-target software based system which helps track the ongoings inside the operation of applications, third-party packages, and the kernel in real time.
Software based tracing is a complex process which requires components on both the target and the host to work together. However its portability across all architectures and board support packages makes it a useful asset. A key requirement in RTEMS trace process is to take existing code in compiled format (ELF) and instrument it in order to log various events and records in real time. However instrumenting of the code for tracing should happen without rebuilding the code from the source and without annotating the source with trace code.
10.1. Introduction to Tracing¶
Tracing is an important function which has several applications including identification of complex threading, detection of deadlocks, tracing functions along with their argument values, and return values through progression of several function calls and audit the performance of an application according to required specifications.
RTEMS tracing framework is under development and welcomes contribution by users.
RTEMS has the following trace components:
Common Trace Format Integration
RTEMS trace framework can currently function using the following methods. Both of the methods make use of the Trace Linker :
10.1.1. RTEMS Trace Using Trace Buffering¶
This scheme of tracing goes through the flow of events described in a subsequent flowchart:
Step 1: The user creates an application and user configuration file. The configuration file specifies the use of the trace buffer generator and other standard initializations. The user then configures their BSP and invokes the trace linker using a command to link the application executable. The trace linker uses the application files in compiled format (ELF) and the libraries used to build the application for performing this link.
Step 2: The RTEMS Trace Linker reads the user’s configuration file and that results in it reading the standard Trace Buffering Configuration files installed with the RTEMS Trace Linker. The trace linker uses the target compiler and linker to create the trace enabled application executable. It wraps the functions defined in the user’s configuration with code that captures trace records into the statically allocated buffer. The trace wrapper code is compiled with the target compiler and the resulting ELF object file is added to the standard link command line used to link the application and the application is re-linked using the wrapping option of the GNU linker.
Step 3: The trace linker creates an executable which is capable of running on the target hardware or simulator.
Step 4: RTEMS shell provides the “rtrace” command to display and save trace buffers.
10.1.2. RTEMS Trace Using Printk¶
This scheme of tracing goes through the flow of events described in a subsequent flowchart:
Step 1: The user creates an RTEMS application in the normal manner as well as a Trace Linker configuration file. The configuration file specifies using the Printk trace mode and the functions to trace. The user invokes the Trace Linker with the configuration and the normal link command line used to the link the application executable. The application ELF object files and libraries, including the RTEMS libraries are standard and do not need to be built specially.
Step 2: The RTEMS Trace Linker reads the user’s configuration file and that results in it reading the standard Printk Trace Configuration files installed with the RTEMS Trace Linker. The trace linker uses the target compiler and linker to create the trace enabled application executable. It wraps the functions defined in the user’s configuration with code that prints the entry with arguments and exit and return value if any. The trace wrapper code is compiled with the target compiler and the resulting ELF object file is added to the standard link command line used to link the application and the application is relinked using the wrapping option of the GNU linker.
Step 3: The trace linker creates and RTEMS ELF executable that can be run on the target hardware or simulator.
Step 4: The application is run in the hardware directly or using a debugger. The printk() output appears on the target console and the user can save that to a file.
The Tracing Examples section describes generation of traces using Trace Buffering technique for the fileio testsuite available with RTEMS installation.
10.2. Tracing Examples¶
The following example executes RTEMS trace using trace buffering for the fileio sample testcase.
10.2.1. Features¶
Tracing using trace buffering consists of the following sets of features:
Individual entry and exit records.
Task details such as CPU, current priority, real priority, task state and interrupt state.
Nano-second timestamp.
Interrupt safe buffer management.
Function argument capture.
Return value capture.
Shell command support to report to the console, save a buffer, assess status of tracing, or view buffers between specified index ranges.
10.2.2. Prerequisites¶
Setup RTEMS for the sparc/erc32 architecture-bsp pair to run the following example.
Download the fileio configuration file and store it on the top of the installed BSP’s directory.
Change the value of the keys: rtems-path and prefix according to your rtems installation. The rtems-path is the path to the bsp installation and prefix is the path to the tools used to build rtems. Also set the value of the rtems-bsp key to sparc/erc32.
10.2.3. Demonstration¶
Inside the RTEMS build directory (the directory where the fileio configuration has been stored) run the following commands to generate traces:
BSP is configured with the following command -
../rtems/configure --target=sparc-rtems5 --prefix=/development/rtems/5 \
--enable-networking --enable-tests --enable-rtemsbsp=erc32 --enable-cxx
The next two commands are used to link the fileio executable.The -B option signifies the use of the complete path to the required directory or file. Write the full path instead of the path file: sparc-rtems5/erc32/lib/ in the following commands according to your installation. Also confirm the path of the fileio’s executable and object files in the last line of the command according to your installation.
sparc-rtems5-gcc -Bsparc-rtems5/erc32/lib/ \
-specs bsp_specs -qrtems -mcpu=cypress -O2 -g -ffunction-sections \
-fdata-sections -Wall -Wmissing-prototypes -Wimplicit-function-declaration \
-Wstrict-prototypes -Wnested-externs -Wl,--gc-sections -mcpu=cypress \
-o sparc-rtems5/c/erc32/testsuites/samples/fileio.exe sparc-rtems5/c/erc32/\
testsuites/samples/fileio/fileio-init.o
This is the trace linker command to generate and compile the wrapper c file for the application. The link command follows the escape sequence “–”. “-C” option denotes the name of the user configuration file and “-W” specifies the name of the wrapper c file.
rtems-tld -C fileio-trace.ini -W fileio-wrapper -- -Bsparc-rtems5/erc32/lib/ \
-specs bsp_specs -qrtems -mcpu=cypress -O2 -g -ffunction-sections \
-fdata-sections -Wall -Wmissing-prototypes -Wimplicit-function-declaration \
-Wstrict-prototypes -Wnested-externs -Wl,--gc-sections -mcpu=cypress \
-o sparc-rtems5/c/erc32/testsuites/samples/fileio.exe sparc-rtems5/c/erc32/\
testsuites/samples/fileio/fileio-init.o
The following command is used to run the application. Hit enter key quickly and type “s” and “root” and “pwd” to run the rtems shell. Use the rtrace status, rtrace trace and rtrace save commands to know the status of the tracing, display the contents of the trace buffer and save the buffer to disk in the form of binary files. Use rtrace -l to list the availalble options for commands with rtrace.
sparc-rtems5-run sparc-rtems5/c/erc32/testsuites/samples/fileio.exe
The output from the above commands will be as follows:
*** BEGIN OF TEST FILE I/O ***
*** TEST VERSION: 5.0.0.de9b7d712bf5da6593386fd4fbca0d5f8b8431d8
*** TEST STATE: USER_INPUT
*** TEST BUILD: RTEMS_NETWORKING RTEMS_POSIX_API
*** TEST TOOLS: 7.3.0 20180125 (RTEMS 5, RSB a3a6c34c150a357e57769a26a460c475e188438f, Newlib 3.0.0)
Press any key to start file I/O sample (20s remaining)
Press any key to start file I/O sample (19s remaining)
Press any key to start file I/O sample (18s remaining)
Press any key to start file I/O sample (17s remaining)
Press any key to start file I/O sample (16s remaining)
Press any key to start file I/O sample (15s remaining)
Press any key to start file I/O sample (14s remaining)
=========================
RTEMS FILE I/O Test Menu
=========================
p -> part_table_initialize
f -> mount all disks in fs_table
l -> list file
r -> read file
w -> write file
s -> start shell
Enter your selection ==>s
Creating /etc/passwd and group with four useable accounts:
root/pwd
test/pwd
rtems/NO PASSWORD
chroot/NO PASSWORD
Only the root user has access to all available commands.
=========================
starting shell
=========================
Welcome to rtems-5.0.0 (SPARC/w/FPU/erc32)
COPYRIGHT (c) 1989-2008.
On-Line Applications Research Corporation (OAR).
Login into RTEMS
/dev/foobar login: root
Password:
RTEMS Shell on /dev/foobar. Use 'help' to list commands.
SHLL [/] # rtrace status
RTEMS Trace Bufferring: status
Running: yes
Triggered: yes
Level: 0%
Traces: 25
SHLL [/] # rtrace stop
RTEMS Trace Bufferring: stop
SHLL [/] # rtrace trace
RTEMS Trace Bufferring: trace
Trace buffer: 0x20921d8
Words traced: 1487
Traces: 25
0:00:40.983197010 2081910 0a010002 [ 2/ 2] > malloc((size_t) 00000130)
0:00:40.983333119 136109 0a010002 [ 2/ 2] < malloc => (void*) 0x219bb88
0:00:40.983471669 138550 0a010002 [ 2/ 2] > malloc((size_t) 00000006)
0:00:40.983606557 134888 0a010002 [ 2/ 2] < malloc => (void*) 0x219bcc0
0:00:40.983684682 78125 0a010002 [ 2/ 2] > malloc((size_t) 00000007)
0:00:40.983819569 134887 0a010002 [ 2/ 2] < malloc => (void*) 0x219bcd0
0:00:40.983909901 90332 0a010002 [ 2/ 2] > malloc((size_t) 000003fc)
0:00:40.984046620 136719 0a010002 [ 2/ 2] < malloc => (void*) 0x219bce0
0:00:40.986624137 2577517 0a010003 [200/200] > malloc((size_t) 00000080)
0:00:40.986767569 143432 0a010003 [200/200] < malloc => (void*) 0x219bce0
0:00:40.987531119 763550 0a010003 [200/200] > calloc((size_t) 00000001, (size_t) 0000005d)
0:00:40.987603751 72632 0a010003 [200/200] > malloc((size_t) 0000005d)
0:00:40.987744743 140992 0a010003 [200/200] < malloc => (void*) 0x219bce0
0:00:40.987824699 79956 0a010003 [200/200] < calloc => (void*) 0x219bce0
0:00:40.988302604 477905 0a010003 [200/200] > malloc((size_t) 00000080)
0:00:40.988446647 144043 0a010003 [200/200] < malloc => (void*) 0x219bd48
0:00:40.988667595 220948 0a010003 [200/200] > calloc((size_t) 00000001, (size_t) 00000080)
0:00:40.988740837 73242 0a010003 [200/200] > malloc((size_t) 00000080)
0:00:40.988884880 144043 0a010003 [200/200] < malloc => (void*) 0x219bdd0
0:00:40.988964836 79956 0a010003 [200/200] < calloc => (void*) 0x219bdd0
0:00:40.989042961 78125 0a010003 [200/200] > calloc((size_t) 00000001, (size_t) 00000080)
0:00:40.989110100 67139 0a010003 [200/200] > malloc((size_t) 00000080)
0:00:40.989254143 144043 0a010003 [200/200] < malloc => (void*) 0x219be58
0:00:40.989334099 79956 0a010003 [200/200] < calloc => (void*) 0x219be58
0:00:40.990118401 784302 0a010003 [200/200] > calloc((size_t) 00000001, (size_t) 00000061)
0:00:40.990176995 58594 0a010003 [200/200] > malloc((size_t) 00000061)
0:00:40.990309441 132446 0a010003 [200/200] < malloc => (void*) 0x219bd48
0:00:40.990384515 75074 0a010003 [200/200] < calloc => (void*) 0x219bd48
0:00:40.990870355 485840 0a010003 [200/200] > malloc((size_t) 00000080)
0:00:40.991011346 140991 0a010003 [200/200] < malloc => (void*) 0x219bee0
0:00:40.991227411 216065 0a010003 [200/200] > calloc((size_t) 00000001, (size_t) 00000080)
0:00:40.991296380 68969 0a010003 [200/200] > malloc((size_t) 00000080)
0:00:40.991438593 142213 0a010003 [200/200] < malloc => (void*) 0x219bf68
0:00:40.991514276 75683 0a010003 [200/200] < calloc => (void*) 0x219bf68
0:00:40.991589349 75073 0a010003 [200/200] > calloc((size_t) 00000001, (size_t) 00000080)
0:00:40.991653437 64088 0a010003 [200/200] > malloc((size_t) 00000080)
0:00:40.991794428 140991 0a010003 [200/200] < malloc => (void*) 0x219bff0
0:00:40.991871332 76904 0a010003 [200/200] < calloc => (void*) 0x219bff0
0:00:40.992283320 411988 0a010003 [200/200] > malloc((size_t) 00000008)
SHLL [/] # rtrace save fileio-trace.bin
RTEMS Trace Bufferring: trace
Trace File: fileio-trace.bin
Trace buffer: 0x20921d8
Words traced: 1487
Traces: 25
SHLL [/] #
10.3. Capture Engine¶
Capture Engine is a trace tool built inside the RTEMS operating system. Capture Engine is designed to cause the lowest load on the system when operating. Hence it does not effect RTEMS when operating or when disabled. It binds to RTEMS at runtime and does not require RTEMS or your application to be rebuilt in order to use it.
The Capture Engine’s sample testcase for the sparc/erc32 is available in build directory created when building RTEMS in the path file: sparc-rtems5/c/erc32/testsuites/samples. In order to access the capture testcase perform the following set of operations inside the RTEMS build directory.
$ cd /sparc-rtems5/c/erc32/testsuites/samples
$ sparc-rtems5-run ./capture.exe
*** BEGIN OF TEST CAPTURE ENGINE ***
*** TEST VERSION: 5.0.0.de9b7d712bf5da6593386fd4fbca0d5f8b8431d8
*** TEST STATE: USER_INPUT
*** TEST BUILD: RTEMS_NETWORKING RTEMS_POSIX_API
*** TEST TOOLS: 7.3.0 20180125 (RTEMS 5, RSB a3a6c34c150a357e57769a26a460c475e188438f, Newlib 3.0.0)
Press any key to start capture engine (20s remaining)
Press any key to start capture engine (19s remaining)
Press any key to start capture engine (18s remaining)
Monitor ready, press enter to login.
1-rtems $
Capture Engine comes with a set of commands to perform various actions.
10.3.1. Capture Engine Commands¶
copen <buffer-size>: Used to initialize the Capture Engine with the trace buffer size in bytes. By default the Capture Engine is not initialized and not running.cwceil <priority-value>: Capture Engine filter used to put an upper limit on the event priority to be captured.cwfloor <priority-value>: Capture Engine filter used to put a lower limit on the event priority to be captured.cwglob <on/off>: Enable or disable the global watch.cenable: Enables the Capture Engine. Capture Engine is by default disabled after being opened.cdisable: Disables the Capture Engine.ctlist: Lists the watch and trigger configurations.ctrace: Dumps the recorded traces. By default this command displays 24 trace records. Repeated use of this command will display all the recorded traces.cwadd <task-name>: Add watch on a particular task.cwtctl <task-name> <on/off>: Enable or disable watch on a particular task.ctset: Used to set a trigger. The general form of the command is:
ctset [-?] type [to name/id] [from] [from name/id]
type in the above command refers to the type of trigger needed. The types of triggers that currently exist are:
switch : a context switch from one task to another task
create : the executing task creates a task
start : the executing task starts a task
restart : the executing task restarts a task
delete : the executing task deletes a task
begin : a task is beginning
exitted : a task is exitting
10.3.2. Example¶
The following is a sample run of the capture testsuite. The test1 command on the Capture Engine Command Line Interface (CLI) makes the RMON task invoke a call to the capture_test_1() command. This function (in the test1.c source code) creates and starts three tasks : CT1a, CT1b and CT1c. These tasks are passed the object id of a semaphore as a task argument. This run through traces the context switches between these tasks. cwceil and cwfloor are set to a narrow range of task priorities to avoid creating noise from a large number of context switches between tasks we are not interested in.
*** BEGIN OF TEST CAPTURE ENGINE ***
*** TEST VERSION: 5.0.0.de9b7d712bf5da6593386fd4fbca0d5f8b8431d8
*** TEST STATE: USER_INPUT
*** TEST BUILD: RTEMS_NETWORKING RTEMS_POSIX_API
*** TEST TOOLS: 7.3.0 20180125 (RTEMS 5, RSB a3a6c34c150a357e57769a26a460c475e188438f, Newlib 3.0.0)
Press any key to start capture engine (20s remaining)
Press any key to start capture engine (19s remaining)
Press any key to start capture engine (18s remaining)
Press any key to start capture engine (17s remaining)
Monitor ready, press enter to login.
1-rtems $ copen 50000
capture engine opened.
1-rtems $ cwceil 100
watch ceiling is 100.
1-rtems $ cwfloor 102
watch floor is 102.
1-rtems $ cwglob on
global watch enabled.
1-rtems $ ctset RMON
trigger set.
1-rtems $ cenable
capture engine enabled.
1-rtems $ test1
1-rtems $ cdisable
capture engine disabled.
1-rtems $ ctrace
0 0:18:17.462314124 0a010003 CT1a 102 102 102 4096 TASK_RECORD
0 0:18:17.462398963 0 0a010003 CT1a 102 102 CREATED
0 0:18:17.462647987 249024 0a010003 CT1a 102 102 STARTED
0 0:18:17.462904334 256347 0a010003 CT1a 102 102 SWITCHED_IN
0 0:18:17.463069129 164795 0a010003 CT1a 102 102 BEGIN
0 0:18:17.463335853 266724 0a010003 CT1a 102 102 SWITCHED_OUT
0 0:18:18.461348547 0a010004 CT1b 101 101 101 4096 TASK_RECORD
0 0:18:18.461433997 998098144 0a010004 CT1b 101 101 CREATED
0 0:18:18.461683631 249634 0a010004 CT1b 101 101 STARTED
0 0:18:18.461934485 250854 0a010004 CT1b 101 101 SWITCHED_IN
0 0:18:18.462099891 165406 0a010004 CT1b 101 101 BEGIN
0 0:18:19.460935339 998835448 0a010004 CT1b 101 101 SWITCHED_OUT
0 0:18:19.461431555 0a010005 CT1c 100 100 100 4096 TASK_RECORD
0 0:18:19.461516394 581055 0a010005 CT1c 100 100 CREATED
0 0:18:19.461765418 249024 0a010005 CT1c 100 100 STARTED
0 0:18:19.462019324 253906 0a010005 CT1c 100 100 SWITCHED_IN
0 0:18:19.462184119 164795 0a010005 CT1c 100 100 BEGIN
0 0:18:19.462475257 291138 0a010005 CT1c 100 100 SWITCHED_OUT
0 0:18:19.462551551 76294 0a010004 CT1b 101 101 SWITCHED_IN
0 0:18:19.960935645 498384094 0a010004 CT1b 101 101 SWITCHED_OUT
0 0:18:19.961012549 76904 0a010003 CT1a 102 100 SWITCHED_IN
0 0:18:19.961341528 328979 0a010003 CT1a 102 102 SWITCHED_OUT
1-rtems $ ctrace
0 0:18:19.961418433 0 0a010005 CT1c 100 100 SWITCHED_IN
0 0:18:19.961672339 253906 0a010005 CT1c 100 100 SWITCHED_OUT
0 0:18:19.961749854 77515 0a010004 CT1b 101 101 SWITCHED_IN
0 0:18:20.460967077 499217223 0a010004 CT1b 101 101 SWITCHED_OUT
0 0:18:20.461219763 252686 0a010005 CT1c 100 100 SWITCHED_IN
0 0:18:20.461424231 204468 0a010005 CT1c 100 100 TERMINATED
0 0:18:20.461747107 322876 0a010005 CT1c 100 100 SWITCHED_OUT
0 0:18:20.461824011 76904 0a010004 CT1b 101 101 SWITCHED_IN
0 0:18:20.462015052 191041 0a010004 CT1b 101 101 TERMINATED
0 0:18:20.462336707 321655 0a010004 CT1b 101 101 SWITCHED_OUT
0 0:18:20.462414222 77515 0a010003 CT1a 102 102 SWITCHED_IN
0 0:18:20.462608924 194702 0a010003 CT1a 102 102 TERMINATED
0 0:18:20.462933021 324097 0a010003 CT1a 102 102 SWITCHED_OUT
1-rtems $ ctrace
1-rtems $
10.4. Trace Linker¶
RTEMS trace linker is a post link tool central to the RTEMS trace framework. It is installed as a part of the RTEMS Tool Project. The RTEMS Trace Linker is a post link tool that performs a re-link of your application to produce a trace executable. A trace executable has been instrumented by the RTEMS Trace Linker with additional code that implements software tracing. A key requirement of the trace process in RTEMS is to take existing code in a compiled format (ELF) and instrument it without rebuilding that code from source and without annotating that source with trace code.
10.4.1. Command Line¶
A typical command to invoke the trace linker consists of two parts separated by --. The first part controls the trace linker and provides the various options it needs and the second part is a standard linker command line you would use to link an RTEMS application. The current command line for trace linker consists of:
$ rtems-tld -h
rtems-trace-ld [options] objects
Options and arguments:
-h : help (also --help)
-V : print linker version number and exit (also --version)
-v : verbose (trace import parts), can supply multiple times
to increase verbosity (also --verbose)
-w : generate warnings (also --warn)
-k : keep temporary files (also --keep)
-c compiler : target compiler is not standard (also --compiler)
-l linker : target linker is not standard (also --linker)
-E prefix : the RTEMS tool prefix (also --exec-prefix)
-f cflags : C compiler flags (also --cflags)
-r path : RTEMS path (also --rtems)
-B bsp : RTEMS arch/bsp (also --rtems-bsp)
-W wrapper : wrapper file name without ext (also --wrapper)
-C ini : user configuration INI file (also --config)
-P path : user configuration INI file search path (also --path)
The trace linker generates code that needs to be compiled and linked to the application executable so it needs to know the target compiler and CFLAGS. There are a couple of ways to do this. The simplest is to provide the path to RTEMS using the -r option and the architecture and BSP name in the standard RTEMS format of arch/bsp. The trace linker will extract the compiler and flags used to build RTEMS and will use them. If you require specific options you can use the -f, -c, -l, and -E options to provide them. If the functions you are tracing use types from your code then add the include path to the CFLAGS.
The trace linker requires you to provide a user configuration file using the -C or --config option. This is an INI format file detailed in the Configuration section. You can also provide an INI file search path using the -P option.
If you are working with new configuration files and you want to view the files the trace linker generates, add the -k option to keep the temporary files, and -W to specify an explicit wrapper C file name. If you set the dump-on-error option in the configuration options section you will get a dump of the configuration on an error.
10.4.2. Configuration (INI) files¶
The Trace Linker is controlled using configuration files. Configuration files are categorized into 3 types:
User Configuration: These are specific to the user application to be traced. This file initializes the values of the trace generator, triggers, enables, and traces.
Tracer Configuration: These are like a library of common or base trace functions that can be referenced by an application. These files tend to hold the details needed to wrap a specific set of functions. Examples provided with the RTEMS Linker are the RTEMS API and Libc.
Generator Configuration: This is used to encapsulate a specific method of tracing. RTEMS currently provides generators for trace buffering, printk, and printf.
The configuration files are in the INI file format which is composed of sections. Each section has a section name and set of keys which consist of names and values. A typical key is of the form name=value. Keys can be used to include other INI files using the include key name. This is shown in the following example where the values indicate rtems and rtld-base configuration files:
include = rtems.ini, rtld-base.ini
The trace linker also uses values in keys to specify other sections. In this example the functions name lists test-trace-funcs and that section contains a headers key that further references a section called test-headers:
functions = test-trace-funcs, rtems-api
[test-trace-funcs]
; Parsed via the 'function-set', not parse as a 'trace'.
headers = test-headers
[test-headers]
header = '#include "test-trace-1.h"'
The format of a configuration file is explained next. Snippets of the file: test-trace.ini have been used for explicit understanding. This file can be found in the rtems-tools directory of the rtems installation.
10.4.2.1. Tracer Section¶
The topmost level section is the tracer section. It can contains the following keys:
name: The name of trace being linked.options: A list of option sections.defines: A list of sections containing defines or define record.define: A list of define string that are single or double quoted.enables: The list of sections containing enabled functions to trace.triggers: The list of sections containing enabled functions to trigger trace on.traces: The list of sections containing function lists to trace.functions: The list of sections containing function details.include: The list of files to include.
The tracer section of the file:test-trace.ini is shown below with explanatory comments.
;
; RTEMS Trace Linker Test Configuration.
;
; We must provide a top level trace section.
;
[tracer]
;
; Name of the trace.
;
name = RTEMS Trace Linker Test
;
; The BSP.
;
bsp = sparc/sis
;
; Functions to trace.
;
traces = test-trace, test-trace-funcs, rtems-api-task
;
; Specify the options.
;
options = test-options
;
; Define the function sets. These are the function's that can be
; added to the trace lists.
;
functions = test-trace-funcs, rtems-api
;
; Include RTEMS Trace support.
;
include = rtems.ini, rtld-base.ini
10.4.2.2. Options section¶
The options section in the fileio-trace.ini is called the fileio-options. A general options section can contain following sets of keys:
dump-on-error: Dump the parsed configuration data on error. The value can be true or false.verbose: Set the verbose level. The value can be true or a number value.prefix: The prefix for the tools and an install RTEMS if rtems-path is not set.cc: The compiler used to compile the generated wrapper code. Overrides the BSP configuration value if a BSP is specified.ld: The linker used to link the application. The default is the cc value as read from the BSP configuration if specified. If your application contains C++ code use this setting to the change the linker to g++.cflags: Set the CFLAGS used to compiler the wrapper. These flags are pre-pended to the BSP read flags if a BSP is specified. This option is used to provide extra include paths to header files in your application that contain types referenced by functions being traced.rtems-path: The path to an install RTEMS if not installed under the prefix.rtems-bsp: The BSP we are building the trace executable for. The is an arch and bsp pair. For example sparc/erc32.
The options section of the file: test-trace.ini uses two of the aforementioned keys as shown below:
;
; Options can be defined here or on the command line.
;
[test-options]
prefix = /development/rtems/5
verbose = true
10.4.2.3. Trace Section¶
A trace section defines how trace wrapper functions are built. To build a trace function that wraps an existing function in an ELF object file or library archive we need to have the function’s signature. A signature is the function’s declaration with any types used. The signature has specific types and we need access to those types which means the wrapper code needs to include header files that define those types. There may also be specific defines needed to access those types. A trace section can contain the following keys:
generator: The generator defines the type of tracing being used.headers: List of sections that contain header file’s keys.header: A header key. Typically the include code.defines: List of sections that contain defines.define: A define key. Typically the define code.signatures: List of function signature sections.trace: Functions that are instrumented with trace code.
The trace section of the file: test-trace.ini is shown below. A trace section can reference other trace sections of a specific type. This allows a trace sections to build on other trace sections.
; User application trace example.
;
[test-trace]
generator = printf-generator
; Just here for testing.
trace = test_trace_3
[test-trace-funcs]
; Parsed via the 'function-set', not parse as a 'trace'.
headers = test-headers
header = '#include "test-trace-2.h"'
defines = test-defines
define = "#define TEST_TRACE_2 2"
signatures = test-signatures
; Parsed via the 'trace', not parsed as a function-set
trace = test_trace_1, test_trace_2
[test-headers]
header = '#include "test-trace-1.h"'
[test-defines]
define = "#define TEST_TRACE_1 1"
[test-signatures]
test_trace_1 = void, int
test_trace_2 = test_type_2, test_type_1
test_trace_3 = float, float*
10.4.2.4. Function Section¶
Function sections define functions that can be traced. Defining a function so it can be traced does not mean it is traced. The function must be added to a trace list to be traced. Function sections provide any required defines, header files, and the function signatures.
A function signature is the function’s declaration. It is the name of the function, the return value, and the arguments. Tracing using function wrappers requires that we have accurate function signatures and ideally we would like to determine the function signature from the data held in ELF files. ELF files can contain DWARF data, the ELF debugging data format. In time the trace project would like to support libdwarf so the DWARF data can be accessed and used to determine a function’s signature. This work is planned but not scheduled to be done and so in the meantime we explicitly define the function signatures in the configuration files.
A function section can consist of the following keys:
headers: A list of sections containing headers or header records.header: A list of include string that are single or double quoted.defines: A list of sections containing defines or define record.defines: A list of define string that are single or double quoted.signatures: A list of section names of function signatures.includes: A list of files to include.
Function signatures are specified with the function name being the key’s name and the key’s value being the return value and a list of function arguments. You need to provide void if the function uses void. Variable argument list are currently not supported. There is no way to determine statically a variable argument list. The function section in the file: test-trace.ini has been labeled as test-trace-funcs. This can be seen in the file snippet of the previous section.
10.4.2.5. Generators¶
The trace linker’s major role is to wrap functions in the existing executable with trace code. The directions on how to wrap application functions is provided by the generator configuration. The wrapping function uses a GNU linker option called –wrap=symbol. The GNU Ld manual states:
“Use a wrapper function for symbol. Any undefined reference to symbol will be resolved to __wrap_symbol. Any undefined reference to __real_symbol will be resolved to symbol.”
Generator sections specify how to generate trace wrapping code. The trace linker and generator section must match to work. The trace linker expects a some things to be present when wrapping functions. The section’s name specifies the generator and can be listed in a generator key in a tracer or trace section. If the generator is not interested in a specific phase it does not need to define it. Nothing will be generated in regard to this phase. For example code to profile specific functions may only provide the entry-trace and exit-trace code where a nano-second time stamp is taken.
The generate code will create an entry and exit call and the generator code block can be used to allocate buffer space for each with the lock held. The entry call and argument copy is performed with the lock released. The buffer space having been allocated will cause the trace events to be in order. The same goes for the exit call. Space is allocated in separate buffer allocate calls so the blocking calls will have the exit event appear in the correct location in the buffer.
The following keys can be a part of the generator configuration:
headers: A list of sections containing headers or header records.header: A list of include string that are single or double quoted.defines: A list of sections containing defines or define record.define: A list of define string that are single or double quoted.entry-trace: The wrapper call made on a function’s entry. Returns bool where true is the function is being traced. This call is made without the lock being held if a lock is defined.arg-trace: The wrapper call made for each argument to the trace function if the function is being traced. This call is made without the lock being held if a lock is defined.exit-trace: The wrapper call made after a function’s exit. Returns bool where true is the function is being traced. This call is made without the lock being held if a lock is defined.ret-trace: The wrapper call made to log the return value if the function is being traced. This call is made without the lock being held if a lock is defined.lock-local: The wrapper code to declare a local lock variable.lock-acquire: The wrapper code to acquire the lock.lock-release: The wrapper code to release the lock.buffer-local: The wrapper code to declare a buffer index local variable.buffer-alloc: The wrapper call made with a lock held if defined to allocate buffer space to hold the trace data. A suitable 32bit buffer index is returned. If there is no space an invalid index is returned. The generator must handle any overhead space needed. The generator needs to make sure the space is available before making the alloc all.code-blocks: A list of code block section names.code: A code block in <<CODE — CODE (without the single quote).includes: A list of files to include.
The following macros can be used in wrapper calls:
@FUNC_NAME@: The trace function name as a quote C string.@FUNC_INDEX@: The trace function index as a held in the sorted list of trace functions by the trace linker. It can be used to index the names, enables, and triggers data.@FUNC_LABEL@: The trace function name as a C label that can be referenced. You can take the address of the label.@FUNC_DATA_SIZE@: The size of the data in bytes.@FUNC_DATA_ENTRY_SIZE@: The size of the entry data in bytes.@FUNC_DATA_RET_SIZE@: The size of the return data in bytes.@ARG_NUM@: The argument number to the trace function.@ARG_TYPE@: The type of the argument as a C string.@ARG_SIZE@: The size of the type of the argument in bytes.@ARG_LABEL@: The argument as a C label that can be referenced.@RET_TYPE@: The type of the return value as a C string.@RET_SIZE@: The size of the type of the return value in bytes.@RET_LABEL@: The return value as a C label that can be referenced.
The buffer-alloc, entry-trace, and exit-trace can be transformed using the following macros:
@FUNC_NAME@@FUNC_INDEX@@FUNC_LABEL@@FUNC_DATA_SZIE@@FUNC_DATA_ENTRY_SZIE@@FUNC_DATA_EXIT_SZIE@
The arg-trace can be transformed using the following macros:
@ARG_NUM@@ARG_TYPE@@ARG_SIZE@@ARG_LABEL@
The ret-trace can be transformed using the following macros:
@RET_TYPE@@RET_SIZE@@RET_LABEL@
The file: test-trace.ini specifies printf-generator as its generator. This section can be found in the file: rtld-print.ini in the rtems-tools directory and is shown below:
;
; A printf generator prints to stdout the trace functions.
;
[printf-generator]
headers = printf-generator-headers
entry-trace = "rtld_pg_printf_entry(@FUNC_NAME@, (void*) &@FUNC_LABEL@);"
arg-trace = "rtld_pg_printf_arg(@ARG_NUM@, @ARG_TYPE@, @ARG_SIZE@, (void*) &@ARG_LABEL@);"
exit-trace = "rtld_pg_printf_exit(@FUNC_NAME@, (void*) &@FUNC_LABEL@);"
ret-trace = "rtld_pg_printf_ret(@RET_TYPE@, @RET_SIZE@, (void*) &@RET_LABEL@);"
code = <<<CODE
static inline void rtld_pg_printf_entry(const char* func_name,
void* func_addr)
{
printf (">>> %s (0x%08x)\n", func_name, func_addr);
}
static inline void rtld_pg_printf_arg(int arg_num,
const char* arg_type,
int arg_size,
void* arg)
{
const unsigned char* p = arg;
int i;
printf (" %2d] %s(%d) = ", arg_num, arg_type, arg_size);
for (i = 0; i < arg_size; ++i, ++p) printf ("%02x", (unsigned int) *p);
printf ("\n");
}
static inline void rtld_pg_printf_exit(const char* func_name,
void* func_addr)
{
printf ("<<< %s (0x%08x)\n", func_name, func_addr);
}
static inline void rtld_pg_printf_ret(const char* ret_type,
int ret_size,
void* ret)
{
const unsigned char* p = ret;
int i;
printf (" rt] %s(%d) = ", ret_type, ret_size);
for (i = 0; i < ret_size; ++i, ++p) printf ("%02x", (unsigned int) *p);
printf ("\n");
}
CODE
[printf-generator-headers]
header = "#include <stdio.h>"
The trace linker generates C code with a wrapper for each function to be instrumented. The trace code generated is driven by the configuration INI files.
10.4.3. Development¶
The Trace Linker is part of the RTEMS tools git repository available at : https://git.rtems.org/rtems-tools The RTEMS tools project utilizes the waf build system. Use the following commands in the topmost build directory to build the tools project:
First we configure using:
$./waf configure --prefix=$HOME/development/rtems/5
Then we build and install using:
$./waf build install
10.5. Event Recording¶
The event recording support focuses on the recording of high frequency events such as
thread switches,
thread queue enqueue and surrender,
interrupt entry and exit,
heap/workspace memory allocate/free,
UMA zone allocate/free,
Ethernet packet input/output, and
etc.
There is a fixed set of 512 system reserved and 512 user defined events which are identified by an event number (rtems_record_event).
The event recording support allows post-mortem analysis in fatal error handlers, e.g. the last events are in the record buffers, the newest event overwrites the oldest event. It is possible to detect record buffer overflows for consumers that expect a continuous stream of events, e.g. to display the system state changes in real-time.
The implementation supports high-end SMP machines (more than 1GHz processor frequency, more than four processors). It uses per-processor ring buffers to record the events. Synchronization is done without atomic read-modify-write operations. The CPU counter is used to get the time of events. It is combined with periodic uptime events to synchronize it with the monotonic system clock (CLOCK_MONOTONIC).
To use the event recording three things come into play. Firstly, there is the generation of event records on the target system (the application running with RTEMS). Secondly, means to transfer the recorded events to the host computer for analysis. Thirdly, the analysis of the recorded events on the host computer.
10.5.1. Target System: Configuration and Event Generation¶
The application enables the event recording support via the configuration option CONFIGURE_RECORD_PER_PROCESSOR_ITEMS. The configuration option CONFIGURE_RECORD_EXTENSIONS_ENABLED enables the generation of thread create, start, restart, delete, switch, begin, exitted and terminate events. Dumps of the event records in a fatal error handler can be enabled by the mutually exclusive CONFIGURE_RECORD_FATAL_DUMP_BASE64 and CONFIGURE_RECORD_FATAL_DUMP_BASE64_ZLIB configuration options.
Custom events can be recorded for example with the rtems_record_produce(), rtems_record_line(), rtems_record_caller(), etc. functions.
#include <rtems/record.h>
void f( void )
{
rtems_record_produce( RTEMS_RECORD_USER( 0 ), 123 );
rtems_record_line();
rtems_record_caller();
}
The variants of rtems_record_line() and rtems_record_caller() can be used to easily generate control flow events in the area of interest. The rtems-record-lttng tool can use these events to associate source code files and line numbers to them using the ELF file of the application.
The following code can be used together with the GCC option -finstrument-functions to generate function entry/exit events for instrumented functions:
__attribute__(( __no_instrument_function__ ))
void __cyg_profile_func_enter( void *this_fn, void *call_site )
{
rtems_record_produce_2(
RTEMS_RECORD_CALLER,
(rtems_record_data) call_site,
RTEMS_RECORD_FUNCTION_ENTRY,
(rtems_record_data) this_fn
);
}
__attribute__(( __no_instrument_function__ ))
void __cyg_profile_func_exit( void *this_fn, void *call_site )
{
rtems_record_produce(
RTEMS_RECORD_FUNCTION_EXIT,
(rtems_record_data) this_fn
);
}
To generate interrupt handler entry/exit events, the following patch can be used:
diff --git a/bsps/arm/shared/clock/clock-armv7m.c b/bsps/arm/shared/clock/clock-armv7m.c
index 255de1ca42..0d37c63ac6 100644
--- a/bsps/arm/shared/clock/clock-armv7m.c
+++ b/bsps/arm/shared/clock/clock-armv7m.c
@@ -29,6 +29,7 @@
#include <bsp/clock-armv7m.h>
#include <rtems.h>
+#include <rtems/record.h>
#include <rtems/sysinit.h>
#ifdef ARM_MULTILIB_ARCH_V7M
@@ -45,9 +46,11 @@ static uint32_t _ARMV7M_TC_get_timecount(struct timecounter *base)
void _ARMV7M_Clock_handler(void)
{
+ rtems_record_produce(RTEMS_RECORD_INTERRUPT_ENTRY, ARMV7M_VECTOR_SYSTICK);
_ARMV7M_Interrupt_service_enter();
Clock_isr(NULL);
_ARMV7M_Interrupt_service_leave();
+ rtems_record_produce(RTEMS_RECORD_INTERRUPT_EXIT, ARMV7M_VECTOR_SYSTICK);
}
static void _ARMV7M_Clock_handler_install(void)
diff --git a/bsps/include/bsp/irq-generic.h b/bsps/include/bsp/irq-generic.h
index 31835d07ba..2ab2f78b65 100644
--- a/bsps/include/bsp/irq-generic.h
+++ b/bsps/include/bsp/irq-generic.h
@@ -30,6 +30,7 @@
#include <stdbool.h>
#include <rtems/irq-extension.h>
+#include <rtems/record.h>
#include <rtems/score/assert.h>
#ifdef RTEMS_SMP
@@ -258,6 +259,7 @@ void bsp_interrupt_vector_disable(rtems_vector_number vector);
*/
static inline void bsp_interrupt_handler_dispatch(rtems_vector_number vector)
{
+ rtems_record_produce(RTEMS_RECORD_INTERRUPT_ENTRY, vector);
if (bsp_interrupt_is_valid_vector(vector)) {
const bsp_interrupt_handler_entry *e =
&bsp_interrupt_handler_table [bsp_interrupt_handler_index(vector)];
@@ -276,6 +278,7 @@ static inline void bsp_interrupt_handler_dispatch(rtems_vector_number vector)
} else {
bsp_interrupt_handler_default(vector);
}
+ rtems_record_produce(RTEMS_RECORD_INTERRUPT_EXIT, vector);
}
/**
10.5.2. Transfer of Event Records to the Host Computer¶
Recorded events can be sent to a host computer with a record server started by rtems_record_start_server() via a TCP connection.
In the fatal error handler, the event records can be dumped via rtems_putc() in Base64 encoding. Optionally, the event records can be compressed via zlib before they are dumped in Base64 encoding. The compression needs roughly 512KiB of statically allocated memory.
10.5.3. Analysis of Event Records on the Host Computer¶
Use the command line tool rtems-record-lttng to get recorded events from the record server running on the target system or from a file to convert the event records into CTF. It can be also used to read the dumps in Base64 encoding generated by the fatal error handler. The tool outputs the event records in the Common Trace Format (CTF) with some extra support for the Linux Trace Toolkit Next Generation (LTTng). This format can be analysed using babeltrace or Eclipse Trace Compass.
For example, to get the event records from the record server running on the target use:
mkdir new-trace
cd new-trace
rtems-record-lttng -e application.exe -H 192.168.188.84 -l 100000
If everything is set up correctly, then the command produces a metadata file and one stream file stream_0, etc. for each processor which generated event records.
$ ls -l
total 120
-rw-r--r-- 1 user group 108339 Apr 11 15:28 metadata
-rw-r--r-- 1 user group 8701 Apr 11 15:28 stream_0
This output in CTF can be used by babeltrace and Eclipse Trace Compass for further analysis, for example:
$ babeltrace .
[07:28:15.909340000] (+?.?????????) RTEMS THREAD_STACK_CURRENT: { cpu_id = 0 }, { data = 0xB10 }
[07:28:15.909340000] (+0.000000000) RTEMS sched_switch: { cpu_id = 0 }, { prev_comm = "UI1 ", prev_tid = 167837697, prev_prio = 0, prev_state = 0, next_comm = "IDLE/0", next_tid = 0, next_prio = 0 }
[07:28:15.909519999] (+0.000179999) RTEMS THREAD_STACK_CURRENT: { cpu_id = 0 }, { data = 0xD68 }
[07:28:15.909519999] (+0.000000000) RTEMS sched_switch: { cpu_id = 0 }, { prev_comm = "IDLE/0", prev_tid = 0, prev_prio = 0, prev_state = 1026, next_comm = "UI1 ", next_tid = 167837697, next_prio = 0 }
[07:28:15.909579999] (+0.000060000) RTEMS THREAD_STACK_CURRENT: { cpu_id = 0 }, { data = 0xB10 }
...
[07:28:15.999940999] (+0.000000000) RTEMS USER_4: { cpu_id = 0 }, { data = 0x4000192C }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_0: { cpu_id = 0 }, { data = 0x0 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_1: { cpu_id = 0 }, { data = 0x1 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_2: { cpu_id = 0 }, { data = 0x2 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_3: { cpu_id = 0 }, { data = 0x3 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_4: { cpu_id = 0 }, { data = 0x4 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_5: { cpu_id = 0 }, { data = 0x5 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_6: { cpu_id = 0 }, { data = 0x6 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_7: { cpu_id = 0 }, { data = 0x7 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_8: { cpu_id = 0 }, { data = 0x8 }
[07:28:15.999940999] (+0.000000000) RTEMS RETURN_9: { cpu_id = 0 }, { data = 0x9 }
[07:28:15.999940999] (+0.000000000) RTEMS ISR_DISABLE: { cpu_id = 0 }, { code = "generate_events at init.c:154" }
[07:28:15.999940999] (+0.000000000) RTEMS ISR_ENABLE: { cpu_id = 0 }, { code = "Init at init.c:181" }
A dump from the fatal error handler looks like this:
*** BEGIN OF RECORDS BASE64 ZLIB ***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*** END OF RECORDS BASE64 ZLIB ***
Copy everything between the *** BEGIN OF RECORDS BASE64 ZLIB *** and the *** END OF RECORDS BASE64 ZLIB *** markers into a file, for example dump.txt. Use this command to convert the event records into the CTF for further analysis:
rtems-record-lttng -e application.exe -b -z dump.txt
11. Changing Versions¶
Follow the sections provide support help when migratiing applications from one version of RTEMS to a new version.
11.1. RTEMS 4.11 to RTEMS 5¶
This section provides helpful information when migrating from RTEMS 4.11 to RTEMS 5.
11.1.1. Application Configuration Options¶
The evaluation of application configuration options in <rtems/confdefs.h> was reworked during the RTEMS 5 development cycle. All options which let the user define data structures were removed, this includes
CONFIGURE_HAS_OWN_CONFIGURATION_TABLE,CONFIGURE_HAS_OWN_BDBUF_TABLE,CONFIGURE_HAS_OWN_DEVICE_DRIVER_TABLE,CONFIGURE_HAS_OWN_FILESYSTEM_TABLE,CONFIGURE_HAS_OWN_INIT_TABLE,CONFIGURE_HAS_OWN_MOUNT_TABLE,CONFIGURE_HAS_OWN_MULTIPROCESSING_TABLE, andCONFIGURE_POSIX_HAS_OWN_INIT_THREAD_TABLE.
The configuration of SMP schedulers changed. For example, RTEMS_SCHEDULER_EDF_SMP() has now only one parameter. Please read section Clustered Scheduler Configuration in the RTEMS Classic API Guide.
A number of configurations options have moved or are obsolete as a result of internal changes in RTEMS. Some of these will produce a warning indicating the new configuration settings you need to define. If you need to run an application on RTEMS 4.11 and RTEMS 5 the following code example shows how to conditionally define the settings. The example is:
#include <rtems.h>
#if __RTEMS_MAJOR__ < 5
#define CONFIGURE_MAXIMUM_FIFOS 10
#define CONFIGURE_MAXIMUM_PIPES 10
#else
#define CONFIGURE_IMFS_ENABLE_MKFIFO
#endif
#define MAX_FILE_DESCRIPTORS 200
#if __RTEMS_MAJOR__ < 5
#define CONFIGURE_LIBIO_MAXIMUM_FILE_DESCRIPTORS MAX_FILE_DESCRIPTORS
#else
#define CONFIGURE_MAXIMUM_FILE_DESCRIPTORS MAX_FILE_DESCRIPTORS
#endif
11.1.2. Clock Manager¶
The directive rtems_clock_get() was removed. See section Transition Advice for the Removed rtems_clock_get() in the RTEMS Classic API Guide for alternatives.
11.1.3. Networking¶
The following code samples provides a simple way to initialise and start networking with the BSD Library’s (libbsd) networking stack. The simplest method to configure the networking stack is to provide a /etc/rc,conf file on your target. If your target has no non-volatile media for a file system create the rc.conf file each time your application starts.
The rc.conf support in libbsd provides a number of needed support settings. We recommend you search for FreeBSD and rc.conf to view the available online documentation that FreeBSD provides.
In this example the network interface is cgem0, replace with your interface name.
static const char* rc_conf =
"# /etc/rc.conf\n" \
"hostname=\"rtems5-libbsd\"\n" \
"ifconfig_cgem0=\"inet 10.1.2.3 netmask 255.255.255.0 rxcsum txcsum\"\n" \
"ifconfig_cgem0_alias0=\"ether 00:80:81:82:83:84\"\n" \
"defaultrouter=\"10.1.2.1\"\n" \
"telnetd_enable=\"YES\"\n";
void start_network(void)
{
FILE *rc;
int r;
/*
* Initialise libbsd.
*/
rtems_bsd_initialize();
/*
* Create the /etc/rc,conf, assume /etc exists.
*/
rc = fopen("/etc/rc.conf", "w");
if (rc_conf == NULL) {
printf("error: cannot create /etc/rc.conf\n");
exit(1);
}
fprintf(rc, rc_conf);
fclose(rc);
/*
* Arguments are timeout and trace
*/
r = rtems_bsd_run_etc_rc_conf(30, false);
if (r < 0) {
printf("error: loading /etc/rc.conf failed: %s\n",strerror(errno));
exit(1);
}
}
11.1.4. Shell Environment¶
To address resource leaks in the RTEMS shell, the management of shell environments changed. This change may break existing code. Here is an example how a broken Telnet shell can be fixed:
static void
telnet_shell( char *name, void *arg )
{
rtems_shell_env_t env;
/* Previous WRONG approach: memset( &env, 0, sizeof( env) ); */
/* Correct way to initialize the shell environment */
rtems_shell_dup_current_env( &env );
env.devname = name;
env.taskname = "TLNT";
env.login_check = NULL;
env.forever = false;
rtems_shell_main_loop( &env );
}
12. Host Tools¶
The RTEMS kernel is developed on host computers cross-compiled and linking the kernel, language runtime libraries, third-party packages and application source code so it can run on target hardware. RTEMS and some of the hardware it support cannot self-host so we need a range of tools to support the wide range of avaliable host computers users wish to develop on. This section details the tools available on the host computers to help support RTEMS users and developers.
12.1. RTEMS Linker¶
The RTEMS Linker is an RTEMS tool to help build RTEMS application that dynamically loader code into a running RTEMS system.
[ NOT COMPLETE ]
12.2. RTEMS Symbols¶
The RTEMS Symbols (rtems-syms) command is an RTEMS tool to generate symbol tables used by the RTEMS Runtime Loader (RTL). The symbol table contains the exported base kernel symbols user code dynamically loaded can reference.
The RTEMS Runtime Loader supports two methods of loading a symbol table, embedded and runtime loading. Embedding the table requires linking the symbol table with the base image and runtime loading loads the table using the dynamic loader when RTEMS is running.
Embedding the symbol table creates self contained images. A target may not have any external media, for example RTEMS tests, or there is a requirement to avoid the management need to match the symbol table with the kernel base image. Embedding the symbol table requires a 2-pass link process making the application’s build system more complicated.
A dynamically loadable symbol table is simpler to create however the symbol table and the kernel base image must match or the behaviour is undefined. There is currently no mechnanisum to ensure the symbol table and the kernel image match The rtems-syms command is run against the base kernel image and the generated symbol table is installed on to the target hardware and loaded before any other modules.
12.2.1. Symbol Table¶
The symbol table is an ELF object file in the target’s ELF format and is built using the target’s RTEMS C compiler. The rtems-syms command searches for the C compller under the prefix this command is installed under or the system path. If the target’s C compiler is not located in either of these paths use the option -c or --cc to specify the path to the compiler.
The rtems-syms command loads the base kernel image’s ELF file and reads the global or public symbols, creates a temporary C file and then compiles it using the target’s RTEMS C compiler. The command automatically detects the architecture from the base kernel image’s ELF file and uses it to create the C compiler’s name. The option -E or --exec-prefix can be used to override the executable prefix used.
It is important to supply suitable C compiler flags (cflags) that match the kernel image’s so the symbol table can be linked or loaded.
12.2.2. 2-Pass Linking¶
2-Pass linking is used to embed a symbol table in a base kernel image. The first link pass is a normal RTEMS kernel link process. The link output is passed to the rtems-syms command and the -e or --embed option is used. The symbol table object file created by rtems-syms is added to the linker command used in the first pass to create the second pass. The address map will change between the first pass and second pass without causing a problem, the symbol table embedded in the second link pass will adjust the symbol addresses to match.
12.2.3. Command¶
rtems-syms [options] kernel
-
-V,--version¶ Display the version information and then exit.
-
-v,--verbose¶ Increase the verbose level by 1. The option can be used more than once to get more detailed trace and debug information.
-
-w,--warn¶ Enable build warnings. This is useful when debugging symbol table generation.
-
-k,--keep¶ Do not delete temporary files on exit, keep them.
-
-e,--embed¶ Create a symbol table that can be embedded in the base kernel image using a 2-pass link process.
-
-S,--symc¶ Specify the symbol’s C source file. The defautl is to use a temporary file name.
-
-o,--output¶ Specify the ELF output file name.
-
-m,--map¶ Create a map file using the provided file name.
-
-C,--cc¶ Specify the C compile executable file name. The file can be absolute and no path is search or relative and the environment’s path is searched.
-
-E,--exec-prefix¶ Specify the RTEMS tool prefix. For example for RTEMS 5 and the SPARC architecture the prefix is
sparc-rtems5.
-
-c,--cflags¶ Specify the C compiler flags used to build the symbol table with. These should be the same or compatible with the flags used to build the RTEMS kernel.
-
-?,-h¶ Reort the usage help.
12.2.4. Examples¶
Create a dynamlically loaded symbol table for the minimum.exe sample program for the i386/pc686 BSP:
$ rtems-syms -o ms.o i386-rtems5/c/pc686/testsuites/samples/minimum/minimum.exe
$ file ms.o
ms.o: ELF 32-bit LSB relocatable, Intel 80386, version 1 (SYSV), not stripped
Run the same command, this time create a map file:
$ rtems-syms -o ms.o -m ms.map i386-rtems5/c/pc686/testsuites/samples/minimum/minimum.exe
$ head -10 ms.map
RTEMS Kernel Symbols Map
kernel: i386-rtems5/c/pc686/testsuites/samples/minimum/minimum.exe
Globals:
No. Index Scope Type SHNDX Address Size Name
0 931 STB_GLOBAL STT_OBJECT 11 0x0012df08 4 BSPBaseBaud (minimum.exe)
1 1124 STB_GLOBAL STT_OBJECT 11 0x0012d894 4 BSPPrintkPort (minimum.exe)
2 836 STB_GLOBAL STT_FUNC 1 0x00104b00 302 BSP_dispatch_isr (minimum.exe)
3 1156 STB_GLOBAL STT_FUNC 1 0x001082d0 92 BSP_install_rtems_shared_irq_handler (minimum.exe)
4 876 STB_GLOBAL STT_FUNC 1 0x00106500 138 BSP_outch (minimum.exe)
Run the same command with a raise verbose level to observe the stages the command performs:
$ rtems-syms -vvv -o ms.o i386-rtems5/c/pc686/testsuites/samples/minimum/minimum.exe
RTEMS Kernel Symbols 5.a72a462adc18
kernel: i386-rtems5/c/pc686/testsuites/samples/minimum/minimum.exe
cache:load-sym: object files: 1
cache:load-sym: symbols: 1043
symbol C file: /tmp/rld--X7paaa.c
symbol O file: ms.o
execute: i386-rtems5-gcc -O2 -c -o ms.o /tmp/rld--X7paaa.c
execute: status: 0
12.3. RTEMS Executable Infomation¶
The RTEMS Executable Information (rtems-exeinfo) command is an RTEMS tool to display some important parts of an RTEMS executable. RTEMS uses ELF as the final linker output and this tool displays useful RTEMS specific information held in the ELF executable. The tool does not replace tools like readelf, rather it focuses on reporting specific information RTEMS builds into the executable.
12.3.1. System Initialisation¶
Linker based system initialisation automatically lets RTEMS only link into the executable the system initialisation code referenced by the user’s application or indirectly by RTEMS. The technique is a varation of the system initialisation process used in the FreeBSD kernel. It is also similar to the process used by the C++ language to run static constructors before entering main and destructors after exit is called.
Linker based system initialisation collects the address of referenced system initialisation functions in specially named sections. The system initialisation function’s address is placed in a variable and the section attribute of the variable is set to a special section name. The linker is instructed via a special linker command file to collect these variables together to create a table. The start-up code in RTRMS loops over the table of addresses and calling each address or system initialisation function. Special section names given to the variables sorts the table placing the functions in a specific order.
A user places a call to an API function in their application and the linker pulls the API code from the RTEMS kernel library adding it to the executing. The API code the linker loads references the variable containing the address of the that API’s system initialisation function. The linker loads the API system initialisation code into the executable to resolve the external refernece created by the variable. If the user does not reference the API the variables is loaded into the executable and no reference to the API system initialisation code is made so it is not linked into the executable.
The design automatically creates a unique system intialisation table for each executable and the code in RTEMS does not change, there is no special build system tricks, or stub libraries.
The RTEMS Execuable Information reports the tables created and you can use this information to debug any initialisation issues.
12.3.2. Command¶
The rtems-exeinfo tool reports RTEMS specific information about the executable. The init and fini tables print the symbol referenced for each table entry and if the symbol is from the C++ language it is demangled.
rtems-exeinfo
-
-V¶ Display the version information and then exit.
-
-v¶ Increase the verbose level by 1. The option can be used more than once to get more detailed trace and debug information.
-
-a¶ Report all types of output data.
-
-I¶ Report the
initor initialisation table.
-
-F¶ Report the
finior finialisation table.
-
-S¶ Report the sections.
-
-?,-h¶ Reort the usage help.
12.3.3. Examples¶
Prints all reports for the hello.exe for the i386/pc686 BSP:
$ rtems-exeinfo -a i386-rtems5/c/pc686/testsuites/samples/hello/hello.exe
RTEMS Executable Info 5.6f5cfada964c
rtems-exeinfo -a i386-rtems5/c/pc686/testsuites/samples/hello/hello.exe
exe: i386-rtems5/c/pc686/testsuites/samples/hello/hello.exe
Sections: 22
-------------- addr: 0x00000000 0x00000000 size: 0 align: 0 relocs: 0
.bss WA------------ addr: 0x00135760 0x0013b300 size: 23456 align: 32 relocs: 0
.comment ---MS--------- addr: 0x00000000 0x00000083 size: 131 align: 1 relocs: 0
.ctors WA------------ addr: 0x0013322c 0x00133234 size: 8 align: 4 relocs: 0
.data WA------------ addr: 0x00133240 0x0013574c size: 9484 align: 32 relocs: 0
.debug_abbrev -------------- addr: 0x00000000 0x0003c5ce size: 247246 align: 1 relocs: 0
.debug_aranges -------------- addr: 0x00000000 0x00003a18 size: 14872 align: 8 relocs: 0
.debug_info -------------- addr: 0x00000000 0x0032496d size: 3295597 align: 1 relocs: 0
.debug_line -------------- addr: 0x00000000 0x0006606b size: 417899 align: 1 relocs: 0
.debug_loc -------------- addr: 0x00000000 0x0003b704 size: 243460 align: 1 relocs: 0
.debug_ranges -------------- addr: 0x00000000 0x00008128 size: 33064 align: 1 relocs: 0
.debug_str ---MS--------- addr: 0x00000000 0x0001a9d7 size: 109015 align: 1 relocs: 0
.dtors WA------------ addr: 0x00133234 0x0013323c size: 8 align: 4 relocs: 0
.eh_frame -A------------ addr: 0x0012b884 0x0013222c size: 27048 align: 4 relocs: 0
.fini -AE----------- addr: 0x00127fdd 0x00127fe5 size: 8 align: 1 relocs: 0
.init -AE----------- addr: 0x00127fd0 0x00127fdd size: 13 align: 1 relocs: 0
.rodata -A------------ addr: 0x00128000 0x0012b884 size: 14468 align: 32 relocs: 0
.rtemsroset WA------------ addr: 0x00127f94 0x00127fd0 size: 60 align: 4 relocs: 0
.shstrtab -------------- addr: 0x00000000 0x000000c6 size: 198 align: 1 relocs: 0
.strtab -------------- addr: 0x00000000 0x000068ca size: 26826 align: 1 relocs: 0
.symtab -------------- addr: 0x00000000 0x00006290 size: 25232 align: 4 relocs: 0
.text WAE----------- addr: 0x00100000 0x00127f91 size: 163729 align: 16 relocs: 0
Init sections: 3
.ctors
0xffffffff RamSize
0x00000000 REG_EFLAGS
.init
0xfd81ebe8 no symbol
0xff86e8ff no symbol
0x00c2ffff no symbol
.rtemsroset
0x00100310 bsp_work_area_initialize
0x00100440 bsp_start_default
0x001160e0 _User_extensions_Handler_initialization
0x0010fe60 rtems_initialize_data_structures
0x0010fcf0 _RTEMS_tasks_Manager_initialization
0x0010f310 _Semaphore_Manager_initialization
0x0010ed90 _POSIX_Keys_Manager_initialization
0x00113af0 _Thread_Create_idle
0x0010c100 rtems_libio_init
0x0010bec0 rtems_filesystem_initialize
0x00100420 bsp_predriver_hook
0x0010bfb0 _Console_simple_Initialize
0x0010ff30 _IO_Initialize_all_drivers
0x0010fc10 _RTEMS_tasks_Initialize_user_tasks_body
0x0010ccb0 rtems_libio_post_driver
Fini sections: 2
.dtors
0xffffffff RamSize
0x00000000 REG_EFLAGS
.fini
0xfd815ee8 no symbol
0x0000c2ff no symbol
The Init section .rtemsroset shows the initialisation call order for the hello.exe sample application. The order is initialise the BSP work area, call the BSP start up, initialise the User extensions, initialise the RTEMS data structures, then call the various Classic API managers that have been linked into the application. Next any POSIX managers are initialisations, in this case the POSIX Keys manager which is used by the thread local storage (TLS) support. Finally the IO and file system is initialise followed by the drivers.
Print the Init section data for the cdtest.exe for the i386/pc686 BSP:
$ rtems-exeinfo -I i386-rtems5/c/pc686/testsuites/samples/cdtest/cdtest.exe
RTEMS Executable Info 5.6f5cfada964c
rtems-exeinfo -I i386-rtems5/c/pc686/testsuites/samples/cdtest/cdtest.exe
exe: i386-rtems5/c/pc686/testsuites/samples/cdtest/cdtest.exe
Init sections: 3
.ctors
0xffffffff RamSize
0x00100ea0 _GLOBAL__sub_I_rtems_test_name
0x001015d0 __gnu_cxx::__freeres()
0x00101df0 __cxxabiv1::__terminate(void (*)())
0x00102ac0 _GLOBAL__sub_I___cxa_get_globals_fast
0x00103260 std::nothrow
0x00000000 REG_EFLAGS
.init
0xfcb3dbe8 no symbol
0xff86e8ff no symbol
0x00c2ffff no symbol
.rtemsroset
0x001112c0 bsp_work_area_initialize
0x001113f0 bsp_start_default
0x001276c0 _User_extensions_Handler_initialization
0x00121260 rtems_initialize_data_structures
0x001210f0 _RTEMS_tasks_Manager_initialization
0x00120710 _Semaphore_Manager_initialization
0x0011ff70 _POSIX_Keys_Manager_initialization
0x001250d0 _Thread_Create_idle
0x0011d220 rtems_libio_init
0x0011cfe0 rtems_filesystem_initialize
0x001113d0 bsp_predriver_hook
0x0011d0d0 _Console_simple_Initialize
0x00121310 _IO_Initialize_all_drivers
0x00121010 _RTEMS_tasks_Initialize_user_tasks_body
0x0011ddd0 rtems_libio_post_driver
The C++ constructor section .ctors shows you the C++ static objects the RTEMS kernel will construct before calling main.
12.4. RTEMS BSP Builder¶
The RTEMS BSP Builder is an RTEMS developer tool to build RTEMS in ways users do not to test changes to RTEMS. RTEMS has large number of architectures, board support packages and configuration options. This tool provides a standard way to test a change.
12.4.1. Developer Workflows¶
There are a number of RTEMS developers each with a different view or expertise in RTEMS. Developers can work in the generic areas such as scheduling, file systems or the shell, or users can become developers adding a new BSP, or even a new port to a new architecture. A common approach for all these developers is to select a BSP and to work with that BSP. Developers working in a generic areas of RTEMS tend to select a BSP that has good simulator support with good debugging such as QEMU, while developers of a new BSP or a new port tend to work on target hardware. This type of development does not check the other architectures, BSP, and build options and a change may change the number of warnings or introduce build errors. It is important for the RTEMS project to have developers fix these issues before pushing the changes to the master repository to avoid breaking the code for other developers. It is best for a developer to resolve as many issues as they work on changes because comming back to a problem often proves difficult.
The RTEMS BSP Builder forms part of a developers workflow where patches are tested before being pushed to the repository.
12.4.2. Build Characteristics¶
Build characteristic are the various parts of a build that can varied changing what is built. RTEMS can vary builds based on:
Architecture
Board Support Package (BSP)
Build options
BSP Options
The BSP Builder provides a template of builds to try and reduce the possble combinations to something manageable. It is not realistic to build all possible combinations on a single machine in reasonible time.
The RTEMS BSP Builder specifies it builds in terms of:
Profiles
Architectures
BSPs
Builds
The RTEMS BSP Builder builds are created by user options that vary these parameters.
12.4.2.1. Profiles¶
A profile is named collection of architectures and board support packages. When the RTEMS BSP Builder is asked to build a specific profile it builds the BSPs in the specified architectures.
The default configuration provides standard profiles for Tiers. They are:
tier-1(default)tier-2tier-3tier-4
The everythings profile allows all BSPs to be built.
12.4.2.2. Builds¶
A build is a list of builds or a build set and each BSP in a profile, architecture of BSP is built with.
The default configuration provides standard builds based around the commonly varied configure options.
The builds are:
all(default)testsstandard, alsono-testsdebugprofilingsmpsmp-debugposixno-posixposix-debugposix-profilingnetworkno-networknetwork-debugsmp-networksmp-network-debug
12.4.2.2.1. All Build¶
The all build is:
debugprofilingsmpsmp-debugposixno-posixposix-debugposix-profilingnetworkno-networknetwork-debugsmp-networksmp-network-debug
A build maps to specific configuration options. The mappings are:
debug
config:base,config:debug
profiling
config:base,config:profiling
smp
config:base,config:smp
smp-debug
config:base,config:smp,config:debug
posix
config:base,config:posix
no-posix
config:base,config:no-posix
posix-debug
config:base,config:posix,config:debug
posix-profiling
config:base,config:posix,config:profiling
network
config:base,config:network
no-network
config:base,config:no-network
network-debug
config:base,config:network,config:debug
smp-network
config:base,config:smp,config:network
smp-network-debug
config:base,config:smp,config:network,config:debug
12.4.3. Build Configurations¶
Build configurations are configure options. These are mapped to the various builds. The configurations are:
base
--target=@ARCH@-rtems@RTEMS_VERSION@--enable-rtemsbsp=@BSP@--prefix=@PREFIX@
tests
--enable-tests
debug
--enable-debug
no-debug
--disable-debug
profiling
--enable-profiling
no-profiling
--disable-profiling
smp
--enable-smp
no-smp
--disable-smp
posix
--enable-posix
no-posix
--disable-posix
network
--enable-networking
no-network
--disable-networking
12.4.4. Performance¶
The RTEMS BSP Builder is designed to extract the maximum performance from your hardware when building RTEMS. The RTEMS build system is based on autoconf, automake and GNU make. Building consists of two phases:
Configuring
Building
The Configuring phase and the start of the Build phase runs autoconf’s configure scripts. These execute as a single linear process and are not run in parallel even if you specify more than one job to make. The configure part of a build is approximately 30% of the total time and higher if building the tests. Performing a single build at a time will not fully utilized a multi-core machine because of the large amount of time the system is idle.
The RTEMS BSP Builder can run more than one build in parallel. A build can also request make run its build with more than one job. The --jobs option lets a user specify the number of build jobs to run at once and the number of make jobs each build runs with. Together these options can fully load a system and can overload a machine.
Tuning the best ratio of buld jobs to make jobs requires running some builds and observing the system’s performance. If the build job count is too low the system will show idle periods and if you have too many build jobs with too many make jobs the system will have too many processing running and the operating system’s overheads in administting too processes at once lowers the overall performance.
A fast eight core machine where the operating system shows sixteen cores can support a build option of --jobs=5/10. The machine will be fully loaded the average build time is around 18 seconds.
The type of build selected effects the optimum jobs option. For example building the tests changes the percentage of time spent configuring copmared to bulding so the make jobs parameter becomes a dominant factor. Lowering the make jobs value avoids having too many active processes running at once.
12.4.5. Command¶
rtems-bsp-builder [options]
-
-?¶ Display a compact help.
-
-h,--help¶ Display the full help.
-
--prefix¶ Prefix to pass to configure when building a BSP.
-
--rtems-tools¶ The path the RTEMS tools such as the C compiler. This option avoid polluting your path. This path is to the tool’s prefix used to build and install the tools and not exact path to an executable.
-
--rtems¶ The path the RTEMS source tree to build.
-
--build-path¶ The path to build the BSP and place the build output. This can be any path and away from your current directory or the RTEMS source code. The storage does not need to be fast like an SSD.
-
--log¶ The log file.
-
--config-report¶ Print a configuration report and exit.
-
--warnings-report¶ Create a warnings report once all builds have finished.
-
--stop-on-error¶ Stop the build on an error. The default is to build all the builds for a profile.
-
--no-clean¶ Do not remove the build once finished. This option lets you inspect the built output. The amount of output can be large and disks can fill with this option.
-
--profiles¶ Build the comma separated list of profiles. The default is
tier-1.
-
--arch¶ A comma separated list of architecures to build using the selected build.
-
--bsp¶ A comma separated list of BSPs to build where a BSP is of the format
arch/bspusing the selected build.
-
--build¶ The build to be used. The default is
all. See--config-reportfor a list of vlaid builds.
-
--jobs¶ The jobs options where the format is
build-jobs/make-jobs. The default is1/num-coreswherenum-coresis the operating system reported number of cores.
-
--dry-run¶ Do not do the actual builds just show what would be built.
12.4.5.1. Examples¶
The following is a tier-1 profile build of all on a machine where all the source and tools are located on fast SSD disks and the build happens on a spinning disk mounted under build. The build uses a development source tree that is bootstrapped and ready to build. The source can have local patches that need to be regression tested:
$ /opt/rtems/5/bin/rtems-bsp-builder --build-path=/build/rtems \
--rtems-tools=/opt/work/rtems/5 \
--rtems=/opt/work/chris/rtems/kernel/rtems.git \
--profiles=tier-1 \
--jobs=5/10
RTEMS Tools Project - RTEMS Kernel BSP Builder, 5.not_released
Profile(s): tier-1
Cleaning: bsp-builds
[ 1/655] arm/altcycv_devkit (debug) Start
[ 1/655] arm/altcycv_devkit (debug) Creating: bsp-builds/arm/altcycv_devkit.debug
[ 2/655] arm/altcycv_devkit (no-posix) Start
[ 2/655] arm/altcycv_devkit (no-posix) Creating: bsp-builds/arm/altcycv_devkit.no-posix
[ 3/655] arm/altcycv_devkit (posix) Start
[ 1/655] arm/altcycv_devkit (debug) Configuring
[ 3/655] arm/altcycv_devkit (posix) Creating: bsp-builds/arm/altcycv_devkit.posix
[ 2/655] arm/altcycv_devkit (no-posix) Configuring
[ 4/655] arm/altcycv_devkit (posix-debug) Start
[ 1/655] arm/altcycv_devkit (debug) Building
[ 3/655] arm/altcycv_devkit (posix) Configuring
[ 4/655] arm/altcycv_devkit (posix-debug) Creating: bsp-builds/arm/altcycv_devkit.posix-debug
[ 2/655] arm/altcycv_devkit (no-posix) Building
[ 5/655] arm/altcycv_devkit (posix-profiling) Start
[ 4/655] arm/altcycv_devkit (posix-debug) Configuring
[ 3/655] arm/altcycv_devkit (posix) Building
....
[654/655] sparc/ngmp (posix-profiling) PASS
[654/655] sparc/ngmp (posix-profiling) Warnings:0 exes:0 objs:0 libs:0
[654/655] sparc/ngmp (posix-profiling) Finished (duration:0:01:49.002189)
[654/655] sparc/ngmp (posix-profiling) Status: Pass: 655 Fail: 0 (configure:0 build:0)
[655/655] sparc/ngmp (profiling) PASS
[655/655] sparc/ngmp (profiling) Warnings:0 exes:0 objs:0 libs:0
[655/655] sparc/ngmp (profiling) Finished (duration:0:01:260.002098)
[655/655] sparc/ngmp (profiling) Status: Pass: 655 Fail: 0 (configure:0 build:0)
[651/655] sparc/ngmp (no-posix) Cleaning: bsp-builds/sparc/ngmp.no-posix
[652/655] sparc/ngmp (posix) Cleaning: bsp-builds/sparc/ngmp.posix
[653/655] sparc/ngmp (posix-debug) Cleaning: bsp-builds/sparc/ngmp.posix-debug
[654/655] sparc/ngmp (posix-profiling) Cleaning: bsp-builds/sparc/ngmp.posix-profiling
[655/655] sparc/ngmp (profiling) Cleaning: bsp-builds/sparc/ngmp.profiling
Total: Warnings:31689 exes:6291 objs:793839 libs:37897
Failures:
No failure(s)
Average BSP Build Time: 0:00:18.165000
Total Time 3:41:48.075006
Passes: 655 Failures: 0
To build a couple of BSPs you are interested in with tests:
$ /opt/rtems/5/bin/rtems-bsp-builder --build-path=/build/rtems \
--rtems-tools=/opt/work/rtems/5 \
--rtems=/opt/work/chris/rtems/kernel/rtems.git \
----log=lpc-log \
--bsp=arm/lpc2362,arm/lpc23xx_tli800 \
--build=tests \
--jobs=5/12
RTEMS Tools Project - RTEMS Kernel BSP Builder, 5.not_released
BSPS(s): arm/lpc2362, arm/lpc23xx_tli800
Cleaning: bsp-builds
[1/2] arm/lpc2362 (tests) Start
[1/2] arm/lpc2362 (tests) Creating: bsp-builds/arm/lpc2362.tests
[2/2] arm/lpc23xx_tli800 (tests) Start
[2/2] arm/lpc23xx_tli800 (tests) Creating: bsp-builds/arm/lpc23xx_tli800.tests
[1/2] arm/lpc2362 (tests) Configuring
[2/2] arm/lpc23xx_tli800 (tests) Configuring
[1/2] arm/lpc2362 (tests) Building
[2/2] arm/lpc23xx_tli800 (tests) Building
[1/2] arm/lpc2362 (tests) FAIL
[1/2] arm/lpc2362 (tests) Warnings:74 exes:58 objs:1645 libs:74
[1/2] arm/lpc2362 (tests) Finished (duration:0:01:31.708252)
[1/2] arm/lpc2362 (tests) Status: Pass: 0 Fail: 2 (configure:0 build:2)
[2/2] arm/lpc23xx_tli800 (tests) FAIL
[2/2] arm/lpc23xx_tli800 (tests) Warnings:74 exes:51 objs:1632 libs:74
[2/2] arm/lpc23xx_tli800 (tests) Finished (duration:0:01:31.747582)
[2/2] arm/lpc23xx_tli800 (tests) Status: Pass: 0 Fail: 2 (configure:0 build:2)
[1/2] arm/lpc2362 (tests) Cleaning: bsp-builds/arm/lpc2362.tests
[2/2] arm/lpc23xx_tli800 (tests) Cleaning: bsp-builds/arm/lpc23xx_tli800.tests
Total: Warnings:74 exes:109 objs:3277 libs:148
Failures:
1 tests arm/lpc2362 build:
configure: /opt/work/chris/rtems/kernel/rtems.git/configure --target\
=arm-rtems5 --enable-rtemsbsp=lpc2362 --prefix=/opt/rtems/5\
--enable-tests
error: ld/collect2:0 error: math.exe section '.rodata' will not fit
in region 'ROM_INT'; region 'ROM_INT' overflowed by 7284 bytes
2 tests arm/lpc23xx_tli800 build:
configure: /opt/work/chris/rtems/kernel/rtems.git/configure --target\
=arm-rtems5 --enable-rtemsbsp=lpc23xx_tli800\
--prefix=/opt/rtems/5 --enable-tests
error: ld/collect2:0 error: math.exe section '.text' will not fit in
region 'ROM_INT'; region 'ROM_INT' overflowed by 13972 bytes
Average BSP Build Time: 0:00:46.658257
Total Time 0:01:33.316514
Passes: 0 Failures: 2
The summary report printed shows both BSP builds failed with the error detail shown. In this case both are linker related errors where the test do not fit into the target’s available resources.
12.5. RTEMS Tester and Run¶
The RTEMS Tester is a test tool that provides a command line interface to run test executables on supported targets. The tool provides back end support for common simulators, debuggers and boot loaders. Board support package (BSP) configurations for RTEMS are provided and can be used to run all the tests in the RTEMS test suite. The tool and its framework is not specific to RTEMS and can be configured to run any suitable application.
RTEMS is an embedded operating system and is cross-compiled on a range of host machines. The executables run on hardware which can vary widely from open source simulators, commercial simulators, debuggers with simulators, debuggers with hardware specific pods and devices, and targets with boot loaders. Testing RTEMS requires that the cross-compiled test executable is transferred to the target hardware, executed, the output captured and returned to the test host where it is analyzed to determine the test result.
Running the RTEMS tests on your target is very important. It provides you with a traceable record showing that your RTEMS version and its tools are working at the level the RTEMS development team expect when releasing RTEMS. Being able to easily run the tests and verify the results is critical in maintaining high standards.
12.5.1. Available BSP testers¶
You can list the available BSP testers with (not all shown):
$ rtems-test --list-bsps
arm920
beagleboardxm
beagleboneblack
jmr3904-run
jmr3904
mcf5235
pc
psim-run
psim
realview_pbx_a9_qemu
sis-run
sis
xilinx_zynq_a9_qemu
xilinx_zynq_a9_qemu_smp
xilinx_zynq_zc706
xilinx_zynq_zc706_qemu
xilinx_zynq_zedboard
Note
The list is growing all the time and if your BSP is not supported we encourage you to add it and submit the configuration back to the project.
Some of the BSPs may appear more than once in the list. These are aliased BSP configurations that may use different back ends. We will show how to use rtems-test command with the erc32 BSP because it is easy to build and use.
12.5.2. Building RTEMS Tests¶
Build the RTEMS Kernel (See RTEMS Kernel) by cloning the repository, running the bootstrap procedure, building and finally installing the kernel. Be sure to enable tests by using --enable-tests option with configure after running bootstrap.
$ ../../rtems.git/configure --target=sparc-rtems5 \
--enable-tests --enable-rtemsbsp=erc32
$ make
Add the -j option to the make command with the number of parallel jobs to run a parallel build (e.g. -j 8).
Building all the tests takes time and it uses more disk so be patient. When make finishes, all the tests will have been built.
Note
Some BSPs may require a post-build process to be run on the RTEMS ELF executable to create an image suitable for execution. This can be built into the configuration script and the tester will perform a pre-test command to convert the executable to a suitable format for your target.
Before running all the tests it is a good idea to run the hello test. The hello test is an RTEMS version of the classic “Hello World” example and running it shows you have a working toolchain and build of RTEMS ready to run the tests. Running the example using SIS:
$ sparc-rtems5-sis sparc-rtems5/c/erc32/testsuites/samples/hello/hello.exe
SIS - SPARC/RISCV instruction simulator 2.20, copyright Jiri Gaisler 2019
Bug-reports to jiri@gaisler.se
ERC32 emulation enabled
Loaded sparc-rtems5/c/erc32/testsuites/samples/hello.exe, entry 0x02000000
sis> run
*** BEGIN OF TEST HELLO WORLD ***
*** TEST VERSION: 5.0.0.c6d8589bb00a9d2a5a094c68c90290df1dc44807
*** TEST STATE: EXPECTED-PASS
*** TEST BUILD: RTEMS_POSIX_API
*** TEST TOOLS: 7.5.0 20191114 (RTEMS 5, RSB 83fa79314dd87c0a8c78fd642b2cea3138be8dd6, Newlib 3e24fbf6f)
Hello World
*** END OF TEST HELLO WORLD ***
*** FATAL ***
fatal source: 0 (INTERNAL_ERROR_CORE)
fatal code: 5 (INTERNAL_ERROR_THREAD_EXITTED)
RTEMS version: 5.0.0.c6d8589bb00a9d2a5a094c68c90290df1dc44807
RTEMS tools: 7.5.0 20191114 (RTEMS 5, RSB 83fa79314dd87c0a8c78fd642b2cea3138be8dd6, Newlib 3e24fbf6f)
executing thread ID: 0x08a010001
executing thread name: UI1
cpu 0 in error mode (tt = 0x101)
116401 02009ae0: 91d02000 ta 0x0
sis> q
The examples can also be run using GDB with SIS as the back end. SIS can be connected to gdb through a network socket using the gdb remote interface.
Either start SIS with -gdb, or issue the gdb command inside SIS, and connect gdb with target remote:1234. The default port is 1234, the port can be changed using the -port option.
Open a terminal and issue the following command:
$ sparc-rtems5-sis -gdb
SIS - SPARC/RISCV instruction simulator 2.20, copyright Jiri Gaisler 2019
Bug-reports to jiri@gaisler.se
ERC32 emulation enabled
gdb: listening on port 1234
Now open another terminal and issue the following command:
$ sparc-rtems5-gdb sparc-rtems5/c/erc32/testsuites/samples/hello/hello.exe
GNU gdb (GDB) 8.3
Copyright (C) 2019 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "--host=x86_64-linux-gnu --target=sparc-rtems5".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from sparc-rtems5/c/erc32/testsuites/samples/hello.exe...
(gdb) target remote:1234
The target remote:1234 will tell gdb to connect to the sis simulator. After this command the output of the first terminal will change to:
$ sparc-rtems5-sis -gdb
SIS - SPARC/RISCV instruction simulator 2.20, copyright Jiri Gaisler 2019
Bug-reports to jiri@gaisler.se
ERC32 emulation enabled
gdb: listening on port 1234 connected
Before running the executable, it must be loaded, this is done using the load command in gdb, and to run it, issue the continue command.
$ sparc-rtems5-gdb sparc-rtems5/c/erc32/testsuites/samples/hello/hello.exe
GNU gdb (GDB) 8.3
Copyright (C) 2019 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "--host=x86_64-linux-gnu --target=sparc-rtems5".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from sparc-rtems5/c/erc32/testsuites/samples/hello.exe...
(gdb) target remote:1234
Remote debugging using :1234
0x00000000 in ?? ()
(gdb) load
Loading section .text, size 0x17170 lma 0x2000000
Loading section .rtemsroset, size 0x40 lma 0x2017170
Loading section .data, size 0x600 lma 0x20181c0
Start address 0x2000000, load size 96176
Transfer rate: 4696 KB/sec, 270 bytes/write.
(gdb) continue
Continuing.
You can see your executable running in the first terminal.
SIS - SPARC/RISCV instruction simulator 2.20, copyright Jiri Gaisler 2019
Bug-reports to jiri@gaisler.se
ERC32 emulation enabled
gdb: listening on port 1235 connected
X2000000,0:#40
*** BEGIN OF TEST HELLO WORLD ***
*** TEST VERSION: 5.0.0.c6d8589bb00a9d2a5a094c68c90290df1dc44807
*** TEST STATE: EXPECTED-PASS
*** TEST BUILD: RTEMS_POSIX_API
*** TEST TOOLS: 7.5.0 20191114 (RTEMS 5, RSB 83fa79314dd87c0a8c78fd642b2cea3138be8dd6, Newlib 3e24fbf6f)
Hello World
*** END OF TEST HELLO WORLD ***
^Csis> q
For more information on the sis simulator refer to this doc: https://gaisler.se/sis/sis.pdf
There are currently close to 500 separate tests and you can run them all with a single RTEMS Tester command.
12.5.3. Running the Tests¶
The rtems-test command line accepts a range of options. These are discussed later in the manual. Command line arguments without a – prefix are test executables or paths to directories. When using a path to a directory, the directories under that path are searched for any file with a .exe extension. This is the default extension for RTEMS executables built within RTEMS. You can pass more than one executable on the command line.
To run the erc32 tests enter the following command from the top of the erc32 BSP build tree:
$ ~/development/rtems/test/rtems-tools.git/tester/rtems-test \
--log=log_erc32_sis \
--rtems-bsp=erc32-sis \
sparc-rtems5/c/erc32/testsuites/samples
RTEMS Testing - Tester, 5.not_released
[ 1/13] p:0 f:0 u:0 e:0 I:0 B:0 t:0 i:0 | sparc/erc32: base_sp.exe
[ 2/13] p:0 f:0 u:0 e:0 I:0 B:0 t:0 i:0 | sparc/erc32: capture.exe
[ 3/13] p:0 f:0 u:0 e:0 I:0 B:0 t:0 i:0 | sparc/erc32: cdtest.exe
[ 4/13] p:0 f:0 u:0 e:0 I:0 B:0 t:0 i:0 | sparc/erc32: fileio.exe
[ 5/13] p:2 f:0 u:0 e:0 I:0 B:0 t:0 i:0 | sparc/erc32: hello.exe
[ 6/13] p:2 f:0 u:0 e:0 I:0 B:0 t:0 i:0 | sparc/erc32: cxx_iostream.exe
[ 8/13] p:2 f:0 u:0 e:0 I:0 B:0 t:2 i:0 | sparc/erc32: minimum.exe
[ 7/13] p:2 f:0 u:0 e:0 I:0 B:0 t:2 i:0 | sparc/erc32: loopback.exe
[ 9/13] p:3 f:0 u:0 e:0 I:0 B:0 t:3 i:0 | sparc/erc32: nsecs.exe
[10/13] p:3 f:0 u:0 e:0 I:0 B:0 t:3 i:0 | sparc/erc32: paranoia.exe
[11/13] p:4 f:0 u:0 e:0 I:0 B:0 t:3 i:0 | sparc/erc32: pppd.exe
[12/13] p:6 f:0 u:0 e:0 I:0 B:0 t:3 i:0 | sparc/erc32: ticker.exe
[13/13] p:6 f:0 u:0 e:0 I:0 B:0 t:3 i:0 | sparc/erc32: unlimited.exe
Passed: 7
Failed: 0
User Input: 0
Expected Fail: 0
Indeterminate: 0
Benchmark: 0
Timeout: 5
Invalid: 1
Total: 13
Average test time: 0:00:27.963000
Testing time : 0:06:03.519012
The output has been shortened so it fits nicely here. Following the order of appearance above, we have the following:
The RTEMS Tester’s test command. In this example we are using an absolute path.
The
--logoption sends the output to a log file. By default only failed tests log the complete output.The
--rtems-bspoption selects the erc32 BSP.The path to the erc32 BSP tests to run. If you add subdirectories to the path specific tests can be run.
The test results so far. See details below.
Overall results of the run. In this run, 13 tests passed, 5 tests timed out and 1 is invalid. The timeouts are probably due to the tests not having enough time to complete. The default timeout is 180 seconds and some of the interrupt tests need more time. The amount of time each test takes depends on the performance of your host CPU when running the simulations.
The average time per test and the total time taken to run all the tests.
Note
If the path to the testsuites was set to sparc-rtems5/c/erc32/testsuites instead of sparc-rtems5/c/erc32/testsuites/samples then all the executables would have been tested and not just those in samples.
Note
Some BSPs require the use of the specific target architecture GDB command (e.g. RTEMS 5 SPARC GDB command is sparc-rtems5-gdb). As this command is part of the RTEMS tools, the --rtems-tools option should be added to the rtems-test command line, e.g. --rtems-tools=$HOME/development/rtems/5. Not every BSP requires this option so you will need to check the specifics of the BSP configuration you are using in order to determine if this option is needed.
An output line is printed for each test that is executed. The rtems-test command by default runs multiple tests in parallel so you will see a number of tests starting quickly and then new tests start as others finish. For example, the output shown above is from an 8-core processor. Thus, the first 8 tests started in parallel and the status shows the order in which they actually started, which is not necessarily sequential, as it happens in the example above where test 8 started before test 7.
Each output line shows information about the current status of the tests. The status reported in each line is the status when the test starts and not the result of that particular test. Thus, a fail, timeout or invalid count changing means a test running before this test failed. The overall status in the end shows that 7 tests passed, no failures, 5 timeouts and 1 invalid test.
Concerning the output of each line, we have the following:
[ 5/13] p:2 f:0 u:0 e:0 I:0 B:0 t:0 i:0 | sparc/erc32: hello.exe
[ 5/13] indicates the test number, in this case test 5 out of 13 tests.
pis the passed test count (2 in this case).fis the failed test count (0 in this case).uis the count for test marked as “user-input” (tests that expect inputfrom the user).
eis the expected-fail count (tests that are expected to fail).Iis the count for tests the results of which are indeterminate.Bis the count for benchmarked tests.tis the timeout test count.iis the invalid test count.sparc/erc32is the architecture and BSP names.hello.exeis the executable name.
The test log records all the tests and results. The logging mode by default only provides the output history if a test fails, times out, or is invalid. The time taken by each test is also recorded.
The tests must complete in a specified period of time or the test is marked as timed out. The default timeout is 3 minutes and can be globally changed using the --timeout command line option. The time required to complete a test can vary. When simulators are run in parallel, the time taken depends on the resources available on the host machine being used. A test per core is the most stable method even though more tests can be run than available cores. If your machine needs longer or you are using a VM you may need to lengthen the timeout.
12.5.4. Test Status¶
Tests can be marked with one of the following:
Pass
Fail
User-input
Expected-fail
Indeterminate
Benchmark
Timeout
Invalid
The RTEMS console or stdout output from the test is needed to determine the result of the test.
12.5.4.1. Pass¶
A test passes if the start and end markers are seen in the test output. The start marker is *** and the end mark is *** END OF TEST. All tests in the RTEMS test suite have these markers.
12.5.4.2. Fail¶
A test fails if the start marker is seen and there is no end marker.
12.5.4.3. User-input¶
A test marked as “user-input” as it expects input from user.
12.5.4.4. Expected-fail¶
A test that is expected to fail.
12.5.4.5. Indeterminate¶
A test the results of which are indeterminate.
12.5.4.6. Benchmark¶
A benchmarked test.
12.5.4.7. Timeout¶
If the test does not complete within the timeout setting the test is marked as having timed out.
12.5.4.8. Invalid¶
If no start marker is seen the test is marked as invalid. If you are testing on real target hardware things can sometimes go wrong and the target may not initialize or respond to the debugger in an expected way.
12.5.5. Logging¶
The following modes of logging are available:
All (
all)Failures (
failures)None (
none)
This mode is controlled using the command line option --log-mode using the values listed above.
12.5.5.1. All¶
The output of all tests is written to the log.
12.5.5.2. Failures¶
The output of the all tests that do not pass is written to the log.
12.5.5.3. None¶
No output is written to the log.
The output is tagged so you can determine where it comes from. The following is the complete output for the In Memory File System test imfs_fslink.exe running on a Coldfire MCF5235 using GDB and a BDM pod:
[ 11/472] p:9 f:0 t:0 i:1 | m68k/mcf5235: imfs_fslink.exe
> gdb: ..../bin/m68k-rtems4.11-gdb -i=mi --nx --quiet ..../imfs_fslink.exe
> Reading symbols from ..../fstests/imfs_fslink/imfs_fslink.exe...
> done.
> target remote | m68k-bdm-gdbserver pipe 003-005
> Remote debugging using | m68k-bdm-gdbserver pipe 003-005
> m68k-bdm: debug module version 0
> m68k-bdm: detected MCF5235
> m68k-bdm: architecture CF5235 connected to 003-005
> m68k-bdm: Coldfire debug module version is 0 (5206(e)/5235/5272/5282)
> Process 003-005 created; pid = 0
> 0x00006200 in ?? ()
> thb *0xffe254c0
> Hardware assisted breakpoint 1 at 0xffe254c0
> continue
> Continuing.
]
]
] External Reset
]
] ColdFire MCF5235 on the BCC
] Firmware v3b.1a.1a (Built on Jul 21 2004 17:31:28)
] Copyright 1995-2004 Freescale Semiconductor, Inc. All Rights Reserved.
]
] Enter 'help' for help.
]
> Temporary breakpoint
> 1, 0xffe254c0 in ?? ()
> load
> Loading section .text, size 0x147e0 lma 0x40000
> Loading section .data, size 0x5d0 lma 0x547e0
> Start address 0x40414, load size 85424
> Transfer rate: 10 KB/sec, 1898 bytes/write.
> b bsp_reset
> Breakpoint 2 at 0x41274: file ..../shared/bspreset_loop.c, line 14.
> continue
> Continuing.
] dBUG>
]
] *** FILE SYSTEM TEST ( IMFS ) ***
] Initializing filesystem IMFS
]
]
] *** LINK TEST ***
] link creates hardlinks
] test if the stat is the same
] chmod and chown
] unlink then stat the file
] *** END OF LINK TEST ***
]
]
] Shutting down filesystem IMFS
] *** END OF FILE SYSTEM TEST ( IMFS ) ***
> Breakpoint
> 2, bsp_reset () at ..../m68k/mcf5235/../../shared/bspreset_loop.c:14
> 14 {
Result: passed Time: 0:00:10.045447
GDB command line (Note: paths with ‘….’ have been shortened)
Lines starting with
>are from GDB’s console.Line starting with
]are from the target’s console.The result with the test time.
12.5.6. Reporting¶
The RTEMS Tester supports output in a machine parsable format. This can be enabled using the options --report-path and --report-format. Currently, JSON output is supported using these options like so: --report-path="report" --report-format=json
This will produce a file report.json that contains output equivalent to the failure logging mode.
12.5.7. Running Tests in Parallel¶
The RTEMS Tester supports parallel execution of tests by default. This only makes sense if the test back end can run in parallel without resulting in resource contention. Simulators are an example of back ends that can run in parallel. A hardware debug tool like a BDM or JTAG pod can manage only a single test at once so the tests need to be run one at a time.
The test framework manages the test jobs and orders the output in the log in test order. Output is held for completed tests until the next test to be reported has finished.
12.5.8. Command Line Help¶
The rtems-test command line accepts a range of options. You can review the available options by using the --help option.
12.6. RTEMS Boot Image¶
The RTEMS Boot Image (rtems-boot-image) command is an RTEMS tool to create disk images suitable for SD cards to boot RTEMS on a range of boards. The supported hosts are:
FreeBSD
Linux
MacOS
The tool captures the specific details for a host operating system to create a bootable disk image as well as capturing the specific detail of the boards that are supported. The tool brings these detail together under a sigle command line interface that is portable across the supported hosts.
The boot image tool can:
Create a disk image to boot an RTEMS executable
Create a disk image to network boot an RTEMS executable
Convert an RTEMS executable into the format a board’s bootloader can load.
The disk images are suitable for booting a range of hardware that have media interfaces, such as an SD card. The default partition type is the Master Boot Record (MRB) and a single root DOS-FS partition is created.
12.6.1. Boot Loaders¶
The boot image tool supports the following boot loaders:
U-boot
12.6.1.1. U-Boot¶
The U-Boot boards supported are:
BeagleBone (arm-ti-am335x_evm)
Zedboard (arm-xilinx-zynq-common)
These boards can be booted with executable and Flat Device Tree (FDT) blobs on disk or view a network if supported by the boards.
The boot image tool can create the following boot configurations for U-Boot:
Executable
A kernel executable is copied in the disk image, loaded by U-Boot and control is passed to the kernel. A reset is performed if the load fails or the kernel returns control to U-Boot.
Executable and FDT
A kernel executable and FDT blob are copied to the disk image, loaded by U-boot and control is passed to the kernel. A reset is performed if the load fails or the kernel returns control to U-Boot.
Network DHCP and Executable
The board’s network interface is initialised, a DHCP request made and a kernel image loaded using TFTP. The loaded kernel is passed control. A reset is performed if the load fails or the kernel returns control to U-Boot.
Network DHCP, Executable and FDT
The board’s network interface is initialised, a DHCP request made and a kernel image loaded using TFTP. The loaded kernel is passed control. A reset is performed if the load fails or the kernel returns control to U-Boot.
The FDT can be installed in and disk image and loaded from it on each boot.
Network Static IP and Executable
The board’s network interface is initialised with a static IP address and a kernel image loaded using TFTP. The loaded kernel is passed control. A reset is performed if the load fails or the kernel returns control to U-Boot.
Network Static IP, Executable and FDT
The board’s network interface is initialised with a static IP address and a kernel image loaded using TFTP. The loaded kernel is passed control. A reset is performed if the load fails or the kernel returns control to U-Boot.
The FDT can be installed in and disk image and loaded from it on each boot.
12.6.2. Hosts¶
The hosts each require specific set up to run the boot image buildier. The tool creates special devices to access the image as a disk and runs file system partitioning and formatting tools. These tools typically require super user or root access. It is not good practice to run commands like this one as root and so the tool dispatches any specific command that needs higher privileges via sudo. If you see a password prompt please enter your password, not a root password if you have one confgiured.
12.6.2.1. FreeBSD¶
Install the sudo package. All commands used are in standard operating system paths and should not require any specific configurations.
12.6.2.2. Linux¶
The loop back kernel module needs to be loaded.
12.6.2.3. MacOS¶
All command used are part of the base OS. No external packages are required.
12.6.3. Configuration¶
The boot image tool is configured by the file rtems-boot.ini that is installed in the
12.6.4. Command¶
The rtems-boot-image tool creates a boot disk image for a specified board. The command line options are:
rtems-boot-image
-
-h,--help¶ Display the command line help.
-
-l,--log¶ Set the log file name. The default is
rtems-log-boot-image.txt.
-
-v,--trace¶ Enable trace or debug logging.
-
-sIMAGE_SIZE,--image-sizeIMAGE_SIZE¶ Set the image size. The size can be in SI units of
k,m, org. The size needs to be something the host’s parition and format tools will accept and it must be large enough to fit the root partition plus any alignments. The default is64m.
-
-FFS_FORMAT,--fs-formatFS_FORMAT¶ Specify type type of format. The supported formats are
fat16andfat32. The default format isfat16.
-
-SFS_SIZE,--fs-sizeFS_SIZE¶ Set the size of the first partition in the disk image. The partition need to be less than the size of the image plus the alignment. The default size is
autowhich will fill the image with the partition.
-
-AFS_ALIGN,--fs-alignFS_ALIGN¶ Set the alignment of the first partition. The default is
1m.
-
-kKERNEL,--kernelKERNEL¶ Optionally provide a kernel image that is copied into the root partition of the disk image and loaded and run when the board boots. The file is an RTEMS executable in the ELF format which is converted to a format the boot loader can load.
-
-dFDT,--fdtFDT¶ Optionally provide a FDT blob that is copied into the root partition of the disk image and loaded when the board boots. If a kernel is provided or a kernel is loaded via a net boot a kernel boot with FDT is executabled. The file is an FDT blob created by the FDT compiler.
-
-fFILE,--fileFILE¶ Optionally provide a file to be copied to the root partition of the disk image. This option can be provided more than once if more than one file needs to be installed.
-
--net-boot¶ Not used and will be removed.
-
--net-boot-dhcp¶ Configure a network boot using DHCP. The kernel will be loaded using TFTP and the file request can be specific by the
--net-boot-fileoption.
-
--net-boot-ipNET_BOOT_IP¶ Configure a network boot using a static IP address. The kernel will be loaded using TFTP and the file request can be specific by the
--net-boot-fileoption. A server IP needs to be specified using the--net-boot-server.
-
--net-boot-fileNET_BOOT_FILE¶ Specify the kernel image file name requested using the TFTP protocol. The default is
rtems.img.
-
--net-boot-fdtNET_BOOT_FDT¶ Optionally specify the file name of a FDT blob loaded using the TFTP protocol. If a net boot FDT file is provide the kernel will be executable with a suitable kernel and FDT boot command.
-
-UCUSTOM_UENV,--custom-uenvCUSTOM_UENV¶ Optionally provide a custom U-boot
uEnv.txtfile that is copied to into the root directory of the root partition of the disk image.
-
-bBOARD,--boardBOARD¶ Specify the board the disk image is built for. The default board is
listwhich lists the available board configurations.
-
--convert-kernel¶ Convert an RTEMS ELF executable into an image file the selected board’s bootloader can load. This option does not create a disk image. The option can be used to create images that can be loaded when network booting.
-
--no-clean¶ If provided the
builddirectory will not be removed after the disk image has been created.
-
-oOUTPUT,--outputOUTPUT¶ The output file name for the image. If the
--convert-kerneloption is used the conversion is written as this file name and if it is not provided the output file is the built disk image.
-
paths[paths ...]¶ The required paths depend on the mix of other options.
If the
--convert-kerneloption is provided a single path to an RTEMS executable file is required. If this option is not provided the number of paths provided determine how they are processed.If a single path a built U-boot directory is provided the board configuration will automatically find and pick up the first and second stage boot loader executables.
If two paths are provided they are paths to the first and second stage boot loader executables. This can be used with loader images they you have not built.
12.6.5. Examples¶
The examples show the output for FreeBSD. It may vary depending on your type of host, how it is configured and what is running.
If the board option is not provided a list of boards is displayed:
$ rtems-boot-image -o sd-card.img u-boot
RTEMS Tools - Boot Image, 5.0.not_released
Board list: bootloaders (1)
u-boot: 2
u-boot-beaglebone
u-boot-zedboard
Create a disk image from a built U-Boot sandbox:
$ rtems-boot-image -o sd-card.img -b u-boot-beaglebone u-boot
RTEMS Tools - Boot Image, 5.0.not_released
Create image: sd-card.img size 64m
Attach image to device: sd-card.img
Password:
Partition device: md0 as MBR
Format: /dev/md0s1 as fat16
Mount: /dev/md0s1
Install: MLO
Install: u-boot.img
Finished
Cleaning up
Create a 32M byte SD card image with the testsuite’s Hello World executable (hello.exe):
$ rtems-boot-image -o sd-card.img -b u-boot-beaglebone -s 32m -k hello.exe u-boot
RTEMS Tools - Boot Image, 5.0.not_released
Create image: sd-card.img size 32m
Attach image to device: sd-card.img
Password:
Partition device: md0 as MBR
Format: /dev/md0s1 as fat16
Mount: /dev/md0s1
Install: MLO
Install: u-boot.img
Install: hello.exe.img
Uenv template: uenv_exe
Install: uEnv.txt
Finished
Cleaning up
Build the same image using the first and second stage boot loaders:
$ rtems-boot-image -o sd-card.img -b u-boot-beaglebone -s 32m -k hello.exe MLO u-boot.img
RTEMS Tools - Boot Image, 5.0.not_released
Create image: sd-card.img size 32m
Attach image to device: sd-card.img
Password:
Partition device: md0 as MBR
Format: /dev/md0s1 as fat16
Mount: /dev/md0s1
Install: MLO
Install: u-boot.img
Install: hello.exe.img
Uenv template: uenv_exe
Install: uEnv.txt
Finished
Cleaning up
Install and load the TI standard FDT for the Beaglebone Black board with the LibBSD DHCP 01 test application:
$ rtems-boot-image -o sd-card.img -b u-boot-beaglebone -s 32m \
-k dhcpcd01.exe -d am335x-boneblack.dtb MLO u-boot.img
RTEMS Tools - Boot Image, 5.0.not_released
Create image: sd-card.img size 32m
Attach image to device: sd-card.img
Password:
Partition device: md0 as MBR
Format: /dev/md0s1 as fat16
Mount: /dev/md0s1
Install: MLO
Install: u-boot.img
Install: dhcpcd01.exe.img
Install: am335x-boneblack.dtb
Uenv template: uenv_exe_fdt
Install: uEnv.txt
Finished
Cleaning up
Create a DHCP network boot image where the TFTP client requests rtems.img:
$ rtems-boot-image -o sd-card.img -b u-boot-beaglebone -s 32m \
--net-boot-dhcp MLO u-boot.img
RTEMS Tools - Boot Image, 5.0.not_released
Create image: sd-card.img size 32m
Attach image to device: sd-card.img
Password:
Partition device: md0 as MBR
Format: /dev/md0s1 as fat16
Mount: /dev/md0s1
Install: MLO
Install: u-boot.img
Uenv template: uenv_net_dhcp
Install: uEnv.txt
Finished
Cleaning up
Select a specific kernel image to load using TFTP and load a FDT blob from the SD card:
$ rtems-boot-image -o sd-card.img -b u-boot-beaglebone -s 32m \
--net-boot-dhcp --net-boot-file bbb1a.img \
-d am335x-boneblack.dtb MLO u-boot.img
RTEMS Tools - Boot Image, 5.0.not_released
Create image: sd-card.img size 32m
Attach image to device: sd-card.img
Password:
Partition device: md0 as MBR
Format: /dev/md0s1 as fat16
Mount: /dev/md0s1
Install: MLO
Install: u-boot.img
Install: am335x-boneblack.dtb
Uenv template: uenv_net_dhcp
Install: uEnv.txt
Finished
Cleaning up
Create an image where a specific kernel image and FDT blob is loaded using the TFTP protocol:
$ rtems-boot-image -o sd-card.img -b u-boot-beaglebone -s 32m \
--net-boot-dhcp --net-boot-file bbb1a.img \
--net-boot-fdt bbb/am335x-boneblack.dtb MLO u-boot.img
RTEMS Tools - Boot Image, 5.0.not_released
Create image: sd-card.img size 32m
Attach image to device: sd-card.img
Password:
Partition device: md0 as MBR
Format: /dev/md0s1 as fat16
Mount: /dev/md0s1
Install: MLO
Install: u-boot.img
Uenv template: uenv_net_dhcp_net_fdt
Install: uEnv.txt
Finished
Cleaning up
12.7. RTEMS TFTP Proxy¶
The RTEMS TFTP Proxy (rtems-tftp-proxy) command is an RTEMS tool to simplify hardware testing using the RTEMS Test and Run commands. This command lets a test set up support a number of similarly configured boards running tests at the same time by proxying the TFTP session requests. The TFTP and U-Boot section details the process to run a test executable on a network connected board.
The TFTP Proxy approach does not require any special modifications in a boot loader to work and works with any compliant TFTP boot client.
An identical SD card boot configuration can be used in similar board when a test set up has a number of similar boards. There is no need to specialize boot configurations. The TFTP proxy server identifies each board by MAC address.
A configuration file maps a board’s MAC address to a TFTP server’s IP address and port number. This provides a centralized means to partition hardware in a test rack between members of a team, continuous integration services or any other project demands.
The TFTP port number a proixed service runs with does not need to be the privileged TFTP port number removing the need to be root to run the RTEMS Test or Run commands. Only the TFTP Proxy needs to running as a privileged user. The RTEMS Test and Run commands lets you specified the TFTP port to bind too.
12.7.1. Operation¶
A network connected board with a suitable boot loader such as U-Boot is configured to boot using TFTP. The boot loader’s configured TFTP server IP address is the address of the host computer running the TFTP Proxy server or the proxy. The TFTP Proxy runs as root or an administrator as it binds by default to the default TFTP port of 69.
A reset board sends a TFTP read request (RRQ) packet to the host machine running the TFTP proxy on the standard TFTP port (69). The proxy server searches the configuration data for a matching MAC address. A configuration match creates a session, forwarding the read request to the proxied IP and port.
The response from the proxied server identifies the remote session port number and the proxy server knows the board’s client port number from the initial request. The proxy transfers the TFTP data transparently between the session’s ports until the transfer finishes.
An example configuration is three different types of boards used for RTEMS kernel regression testing and application development.
RTEMS TFTP Proxy Test Lab¶
The project has a continuous integration (CI) server on address 10.0.0.100 and two boards, a BeagleBone Black and Xilinx MicroZed board, are confgured for testing. A developer on another host machine is using a RaspberryPi to develop an application. The configuration file is:
;
; Project Foo Test network.
;
[default]
clients = bbb, uzed, rpi2
[bbb]
mac = 1c:ba:8c:96:20:bc
host = 10.0.0.100:9001
[uzed]
mac = 6e:3a:1c:22:aa:5f, 8a:3d:5f:67:55:cb
host = 10.0.0.100:9002
[rpi2]
mac = b8:27:eb:29:6b:bc
host = 10.0.0.110:9001
The MicroZed board can be seen by different MAC addresses depending on how U-Boot starts. As a result both are listed.
12.7.2. Configuration¶
The boot image tool is configured by an INI file that is passed to the TFTP proxy on the command line when it starts.
The [default] section has to contain a clients entry that lists the clients. There needs to be a client section for each listed client.
A client section header is a client name listed in the clients record in the defaults section. A client section has to contain a mac record and a host record. The MAC record is a comma separated list of MAC addresses in the standard 6 octet hex format separated by :. A list of MAC addresses will match for any address listed. The host record is the IP address and port number of the proxied TFTP server.
12.7.3. Command¶
The rtems-tftp-proxy tool runs a TFTP proxy server using a user provided configuration file. The command line options are:
rtems-tftp-proxy
-
-h,--help¶ Display the command line help.
-
-l,--log¶ Set the log file name.
-
-v,--trace¶ Enable trace or debug logging.
-
-cCONFIG,--configCONFIG¶ The INI format configuration file.
-
-BBIND,--bindBIND¶ The interface address the proxy binds too. The default is
allwhich means the proxy binds to all interfaces.
-
-PPORT,--portPORT¶ The port the proxy server binds too. The default is the TFTP standard port of 69. This is a privileged port so if using this port number run the TFTP proxy with root or administrator privileges.
12.7.4. Examples¶
The examples show running the TFTP Proxy as a privileged user:
$ sudo rtems-tftp-proxy -c foo-test-lab.ini
Password:
RTEMS Tools - TFTP Proxy, 5.1.0
Command Line: rtems-tftp-proxy -c foo-test-lab.ini
Host: FreeBSD ruru 12.0-RELEASE-p3 FreeBSD 12.0-RELEASE-p3 GENERIC amd64
Python: 3.6.9 (default, Nov 14 2019, 01:16:50) [GCC 4.2.1 Compatible FreeBSD Clang 6.0.1 (tags/RELEASE_601/final 335540)]
Proxy: all:6
13. Source Builder¶
The RTEMS Source Builder or RSB is a tool to build packages from source. The RTEMS project uses it to build it’s compilers, tools, kernel and 3rd party packages. The tool is aimed at developers of software for RTEMS who use tool sets for embedded development.
The RSB consolidates the details you need to build a package from source in a controlled and verifiable way. The RSB is not limited to building tools for RTEMS, you can build bare metal development environments.
The RSB section of this manual caters for a range of users from new to experienced RTEMS developers who want to understand the RTEMS Source Builder. New users who just want to build tools should follow Quick Start in this manual. Users building a binary tool set for release can read the “Installing and Tar Files”. Users wanting to run and test bleeding edge tools or packages, or wanting update or extend the RSB’s configuration can read the remaining sections.
Embedded development typically uses cross-compiling tool chains, debuggers, and debugging aids. Together we call these a tool set. The RTEMS Source Builder is designed to fit this specific niche but is not limited to it. The RSB can be used outside of the RTEMS project and we welcome this.
The RTEMS Source Builder is typically used to build a set of packages or a build set. A build set is a collection of packages and a package is a specific tool, for example GCC, GDB, or library of code and a single build set can build them all in a single command. The RTEMS Source Builder attempts to support any host environment that runs Python and you can build the package on. The RSB is not some sort of magic that can take any piece of source code and make it build. Someone at some point in time has figured out how to build that package from source and taught this tool.
The RTEMS Source Builder is known to work on:
ArchLinux
CentOS
Fedora
Raspbian
Ubuntu (includes XUbuntu)
Linux Mint
openSUSE
FreeBSD
NetBSD
Solaris
MacOS
Windows
The RTEMS Source Builder has two types of configuration data. The first is the build set. A build set describes a collection of packages that define a set of tools you would use when developing software for RTEMS. For example the basic GNU tool set is Binutils, GCC, and GDB and is the typical base suite of tools you need for an embedded cross-development type project. The second type of configuration data are the configuration files and they define how a package is built. Configuration files are scripts loosely based on the RPM spec file format and they detail the steps needed to build a package. The steps are preparation, building, and installing. Scripts support macros, shell expansion, logic, includes plus many more features useful when build packages.
The RTEMS Source Builder does not interact with any host package management systems. There is no automatic dependence checking between various packages you build or packages and software your host system you may have installed. We assume the build sets and configuration files you are using have been created by developers who do. Support is provided for package config or pkgconfig type files so you can check and use standard libraries if present. If you have a problem please ask on our Developers Mailing List.
Bug Reporting
If you think you have found a problem please see Bugs, Crashes, and Build Failures.
13.1. Why Build from Source?¶
The RTEMS Source Builder is not a replacement for the binary install systems you have with commercial operating systems or open source operating system distributions. Those products and distributions are critically important and are the base that allows the RSB to work. The RTEMS Source Builder sits somewhere between you manually entering the commands to build a tool set and a tool such as yum or apt-get to install binary packages made specifically for your host operating system. Building manually or installing a binary package from a remote repository are valid and real alternatives. The RSB provides the specific service of repeatably being able to build tool sets from source code. The process leaves you with the source code used to build the tools and the ability to rebuild it.
If you are developing a system or product that has a long shelf life or is used in a critical piece of infrastructure that has a long life cycle being able to build from source is important. It insulates the project from the fast ever changing world of the host development machines. If your tool set is binary and you have lost the ability to build it you have lost a degree of control and flexibility open source gives you. Fast moving host environments are fantastic. We have powerful multi-core computers with huge amounts of memory and state of the art operating systems your development uses however the product or project you are part of may need to be maintained well past the life time of these host. Being able to build from source is an important and critical part of this process because you can move to a newer host and create an equivalent tool set.
Building from source provides you with control over the configuration of the package you are building. If all or the most important dependent parts are built from source you limit the exposure to host variations. For example the GNU Compiler Collection (GCC) currently uses a number of third-party libraries internally (GMP, ISL, MPC, MPFR, etc.). If your validated compiler generating code for your target processor is dynamically linked against the host’s version of these libraries any change in the host’s configuration may effect you. The changes the host’s package management system makes may be perfectly reasonable in relation to the distribution being managed however this may not extend to you and your tools. Building your tools from source and controlling the specific version of these dependent parts means you are not exposing yourself to unexpected and often difficult to resolve problems. On the other side you need to make sure your tools build and work with newer versions of the host operating system. Given the stability of standards based libraries like libc and ever improving support for standard header file locations this task is becoming easier.
The RTEMS Source Builder is designed to be audited and incorporated into a project’s verification and validation process. If your project is developing critical applications that needs to be traced from source to executable code in the target, you need to also consider the tools and how to track them.
If your IT department maintains all your computers and you do not have suitable rights to install binary packages, building from source lets you create your own tool set that you install under your home directory. Avoiding installing any extra packages as a super user is always helpful in maintaining a secure computing environment.
13.2. Project Sets¶
The RTEMS Source Builder supports project configurations. Project configurations can be public or private and can be contained in the RTEMS Source Builder project if suitable.
The configuration file loader searches the macro _configdir and by default this is set to %{_topdir}/config:%{_sbdir}/config where _topdir is your current working directory, or the directory you invoke the RTEMS Source Builder command in. The macro _sbdir is the directory where the RTEMS Source Builder command resides. Therefore the config directory under each of these is searched so all you need to do is create a config in your project and add your configuration files. They do not need to be under the RTEMS Source Builder source tree. Public projects are included in the main RTEMS Source Builder such as RTEMS.
You can add your own patches directory next to your config directory as the %patch command searches the _patchdir macro variable and it is by default set to %{_topdir}/patches:%{_sbdir}/patches.
The source-builder/config directory provides generic scripts for building various tools. You can specialise these in your private configurations to make use of them. If you add new generic configurations please contribute them back to the project
Build sets can be controlled via the command line to enable (--with-<feature>) and disable (--without-<feature>) various features. There is no definitive list of build options that can be listed because they are implemented with the configuration scripts. The best way to find what is available is to grep the configuration files for with and without.
13.2.1. Bare Metal¶
The RSB contains a bare configuration tree and you can use this to add packages you use on the hosts. For example ‘qemu’ is supported on a range of hosts. RTEMS tools live in the rtems/config directory tree. RTEMS packages include tools for use on your host computer as well as packages you can build and run on RTEMS.
The bare metal support for GNU Tool chains. An example is the lang/gcc491 build set. You need to provide a target via the command line --target option and this is in the standard 2 or 3 tuple form. For example for an ARM compiler you would use arm-eabi or arm-eabihf, and for SPARC you would use sparc-elf:
$ cd rtems-source-builder/bare
$ ../source-builder/sb-set-builder --log=log_arm_eabihf \
--prefix=$HOME/development/bare --target=arm-eabihf lang/gcc491
RTEMS Source Builder - Set Builder, v0.3.0
Build Set: lang/gcc491
config: devel/expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-apple-darwin13.2.0-1
building: expat-2.1.0-x86_64-apple-darwin13.2.0-1
config: devel/binutils-2.24-1.cfg
package: arm-eabihf-binutils-2.24-1
building: arm-eabihf-binutils-2.24-1
config: devel/gcc-4.9.1-newlib-2.1.0-1.cfg
package: arm-eabihf-gcc-4.9.1-newlib-2.1.0-1
building: arm-eabihf-gcc-4.9.1-newlib-2.1.0-1
config: devel/gdb-7.7-1.cfg
package: arm-eabihf-gdb-7.7-1
building: arm-eabihf-gdb-7.7-1
installing: expat-2.1.0-x86_64-apple-darwin13.2.0-1 -> /Users/chris/development/bare
installing: arm-eabihf-binutils-2.24-1 -> /Users/chris/development/bare
installing: arm-eabihf-gcc-4.9.1-newlib-2.1.0-1 -> /Users/chris/development/bare
installing: arm-eabihf-gdb-7.7-1 -> /Users/chris/development/bare
cleaning: expat-2.1.0-x86_64-apple-darwin13.2.0-1
cleaning: arm-eabihf-binutils-2.24-1
cleaning: arm-eabihf-gcc-4.9.1-newlib-2.1.0-1
cleaning: arm-eabihf-gdb-7.7-1
13.2.2. RTEMS¶
The RTEMS Configurations are found in the rtems directory. The configurations are grouped by RTEMS version and a release normally only contains the configurations for that release.. In RTEMS the tools are specific to a specific version because of variations between Newlib and RTEMS. Restructuring in RTEMS and Newlib sometimes moves libc functionality between these two parts and this makes existing tools incompatible with RTEMS.
RTEMS allows architectures to have different tool versions and patches. The large number of architectures RTEMS supports can make it difficult to get a common stable version of all the packages. An architecture may require a recent GCC because an existing bug has been fixed, however the more recent version may have a bug in other architecture. Architecture specific patches should only be appliaed when build the related architecture. A patch may fix a problem on one architecture however it could introduce a problem in another architecture. Limiting exposure limits any possible crosstalk between architectures.
If you have a configuation issue try adding the --dry-run option. This will run through all the configuration files and if any checks fail you will see this quickly rather than waiting for until the build fails a check.
Following features can be enabled/disabled via the command line for the RTEMS build sets:
--without-cxxDo not build a C++ compiler.
--with-adaAttempt to build an Ada compiler. You need a native GNAT installed.
--with-fortranAttempt to build a Fortran compiler.
--with-objcAttempt to build a C++ compiler.
The RSB provides build sets for some BSPs. These build sets will build:
Compiler, linker, debugger and RTEMS Tools.
RTEMS Kernel for the BSP
Optionally LibBSD if supported by the BSP.
Third party packages if supported by the BSP.
13.2.3. Patches¶
Packages being built by the RSB need patches from time to time and the RSB supports patching upstream packages. The patches are held in a seperate directory called patches relative to the configuration directory you are building. For example %{_topdir}/patches:%{_sbdir}/patches. Patches are declared in the configuration files in a similar manner to the package’s source so please refer to the %source documentation. Patches, like the source, are to be made publically available for configurations that live in the RSB package and are downloaded on demand.
If a package has a patch management tool it is recommended you reference the package’s patch management tools directly. If the RSB does not support the specific patch manage tool please contact the mailing list to see if support can be added.
Referenced patches should be placed in a location that is easy to access and download with a stable URL. We recommend attaching a patch to an RTEMS ticket in it’s bug reporting system or posting to a mailing list with online archives.
RTEMS’s former practice of placing patches in the RTEMS Tools Git repository has been stopped.
Patches are added to a component’s name and in the %prep: section the patches can be set up, meaning they are applied to source. The patches are applied in the order they are added. If there is a dependency make sure you order the patches correctly when you add them. You can add any number of patches and the RSB will handle them efficiently.
Patches can have options. These are added before the patch URL. If no options are provided the patch’s setup default options are used.
Patches can be declared in build set up files.
This examples shows how to declare a patch for gdb in the lm32 architecture:
%patch add <1> gdb <2> %{rtems_gdb_patches}/lm32/gdb-sim-lm32uart.diff <3>
Items:
The patch’s
addcommand.The group of patches this patch belongs too.
The patch’s URL. It is downloaded from here.
Patches require a checksum to avoid a warning. The %hash directive can be used to add a checksum for a patch that is used to verify the patch:
%hash sha512 <1> gdb-sim-lm32uart.diff <2> 77d07087 ... e7db17fb <3>
Items:
The type of checksum, in the case an SHA512 hash.
The patch file the checksum is for.
The SHA512 hash.
The patches are applied when a patch setup command is issued in the %prep: section. All patches in the group are applied. To apply the GDB patch above use:
%patch setup <1> gdb <2> -p1 <3>
Items:
The patch’s
setupcommand.The group of patches to apply.
The patch group’s default options. If no option is given with the patch these options are used.
Architecture specific patches live in the architecture build set file isolating the patch to that specific architecture. If a patch is common to a tool it resides in the RTEMS tools configuration file. Do not place patches for tools in the source-builder/config template configuration files.
To test a patch simply copy it to your local patches directory. The RSB will see the patch is present and will not attempt to download it. Once you are happy with the patch submit it to the project and a core developer will review it and add it to the RTEMS Tools git repository.
13.2.3.1. Testing a Newlib Patch¶
To test a local patch for newlib, you need to add the following two lines to the .cfg file in rsb/rtems/config/tools/ that is included by the bset you use:
Steps:
Create patches for the changes you want to test. (Note: For RSB, before creating Newlib patch, you must run
autoreconf -fviin the required directory after you make changes to the code. This is not required when you create patch to send tonewlib-devel. But if you want RSB to address your changes, your patch should also include regenerated files.)Calculate
sha512of your patch.Place the patches in
rsb/rtems/patchesdirectory.Open the
.bsetfile used by your BSP inrsb/rtems/config. For example, forrtems5,SPARC, the file will bersb/rtems/config/5/rtems-sparc.bset.Inside it you will find the name of
.cfgfile for Newlib, used by your BSP. For example, I foundtools/rtems-gcc-7.4.0-newlib-1d35a003f.Edit your
.cfgfile. In my case it will be,rsb/rtems/config/tools/rtems-gcc-7.4.0-newlib-1d35a003f.cfg. And add the information about your patch as mentioned below.
%patch add newlib -p1 file://0001-Port-ndbm.patch <1>
%hash sha512 0001-Port-ndbm.patch 7d999ceeea4f3dc82e8e0aadc09d983a7a68b44470da8a3d61ab6fc558fdba6f2c2de3acc2f32c0b0b97fcc9ab799c27e87afe046544a69519881f947e7881d1 <2>
Items:
The diff file prepended with
file://to tell RSB this is a local file.The output from sha512sum on the patch file.
13.3. Cross and Canadian Cross Building¶
Cross building and Canadian Cross building is the process of building on one machine an executable that runs on another machine. An example is building a set of RTEMS tools on Linux to run on Windows. The RSB supports cross building and Canadian cross building.
This sections details how to the RSB to cross and Canadian cross build.
13.3.1. Cross Building¶
Cross building is where the build machine and host are different. The build machine runs the RSB and the host machine is where the output from the build runs. An example is building a package such as Curl for RTEMS on your development machine.
To build the Curl package for RTEMS you enter the RSB command:
$ ../source-builder/sb-set-builder \
--log=log_curl_arm.txt \
--prefix=$HOME/development/rtems/5 \ <1>
--host=arm-rtems5 \ <2>
--with-rtems-bsp=xilinx_zynq_zc706 \ <3>
5/ftp/curl
Items:
The tools and the RTEMS BSP are installed under the same prefix.
The
--hostcommand is the RTEMS architecture and version.The BSP is built and installed in the prefix. The architecture must match the
--hostarchitecture.
13.3.2. Canadian Cross Building¶
A Canadian cross builds are where the build, host and target machines all differ. For example building an RTEMS compiler for an ARM processor that runs on Windows is built using a Linux machine. The process is controlled by setting the build triplet to the host you are building, the host triplet to the host the tools will run on and the target to the RTEMS architecture you require. The tools needed by the RSB are:
Build host C and C++ compiler
Host C and C++ cross compiler
The RTEMS Source Builder requires you provide the build host C and C++ compiler and the final host C and C++ cross-compiler. The RSB will build the build host RTEMS compiler and the final host RTEMS C and C++ compiler, the output of this process.
The Host C and C++ compiler is a cross-compiler that builds executables for the host you want the tools for. You need to provide these tools. For Windows a number of Unix operating systems provide MinGW tool sets as packages.
The RSB will build an RTEMS tool set for the build host. This is needed when building the final host’s RTEMS compiler as it needs to build RTEMS runtime code such as libc on the build host.
TIP: Make sure the host’s cross-compiler tools are in your path before run the RSB build command.
TIP: Canadian Cross built tools will not run on the machine being used to build them so you should provide the --bset-tar-files and --no-install options. The option to not install the files lets you provide a prefix that does not exist or you cannot access.
To perform a cross build add --host= to the command line. For example to build a MinGW tool set on FreeBSD for Windows add --host=mingw32 if the cross compiler is mingw32-gcc:
$ ../source-builder/sb-set-builder --host=mingw32 \
--log=l-mingw32-4.11-sparc.txt \
--prefix=$HOME/development/rtems/5 \
5/rtems-sparc
If you are on a Linux Fedora build host with the MinGW packages installed the command line is:
$ ../source-builder/sb-set-builder --host=i686-w64-mingw32 \
--log=l-mingw32-4.11-sparc.txt \
--prefix=$HOME/development/rtems/5 \
5/rtems-sparc
13.4. Third-Party Packages¶
This section describes how to build and add an RTEMS third-party package to the RSB.
A third-party package is a library or software package built to run on RTEMS, examples are Curl, NTP, Net-Snmp, libjpeg and more. These pieces of software can be used to help build RTEMS applications. The package is built for a specific BSP and so requires a working RTEMS tool chain, an installed RTEMS Board Support Package (BSP), and a network stack if the package uses networking resources.
The RSB support for building third-party packages is based around the pkconfig files (PC) installed with the BSP. The pkgconfig support in RTEMS is considered experimental and can have some issues for some BSPs. This issue is rooted deep in the RTEMS build system.
13.4.1. Vertical Integration¶
The RSB supports horizontal integration with support for multiple architectures. Adding packages to the RSB as libraries is vertical integration. Building the GCC tool chain requires you build an assembler before you build a compiler. The same can be done for third-party libraries, you can create build sets that stack library dependences vertically to create a stack.
13.4.2. Building¶
To build a package you need to have a suitable RTEMS tool chain and RTEMS BSP installed. The set builder command line requires you provide the tools path, the RTEMS architecture (host), the BSP, and the prefix path used to the install RTEMS BSP.
The RSB prefix option (--prefix) provided when building a package is the path to:
The tools, RTEMS kernel and any dependent libraries such as LibBSD. The package will be installed into the prefix path. This build configuration can be used to make a complete set of development tools and libraries for a project or product under a single path.
The RTEMS kernel and any dependent libraries such as LibBSD. The tools path needs to be in the environment path (not recommended) or provided to the set builder command by the
--with-toolsoption. The package will be installed into the prefix path. This build configuration can be used when you have a set of tools used with a number of RTEMS BSPs. The tools can be shared between the different BSPs.The path the package is installed into. The tools path needs to be in the environment path (not recommended) or provided to the set builder command using the
--with-toolsoption. The path to the RTEMS kernel and any dependent libraries such as LibBSD needs to be supplied to the set builder command using the--with-rtemsoption. This build configuration can be used when you have a set of libraries you are testing with a changing RTEMS kernel. Becareful using this configuration as changes in RTEMS interfaces may require rebuilding these packages.
The set builder command option --host is used to provide the RTEMS architecture the package is being built for. For example --host=arm-rtems5 is used for any ARM BSP.
The set builder command option --with-rtems-bsp is the RTEMS BSP the package is being built for. The BSP is searched for under the path provided by the command option --with-rtems and if this option is not provided the provided prefix is searched.
The following example builds and installs the Curl networking package for the ARM BeagleBone Black BSP installing it into the same path the tools, RTEMS kernel and LibBSD are installed in.
$ ../source-builder/sb-set-builder --prefix=$HOME/development/cs/rtems/5 \
--log=curl.txt --host=arm-rtems5 --with-rtems-bsp=beagleboneblack ftp/curl
RTEMS Source Builder - Set Builder, 5 (2bdae1f169e4)
Build Set: ftp/curl
config: ftp/curl-7.65.1-1.cfg
package: curl-v7.65.1-arm-rtems5-1
download: https://curl.haxx.se/download/curl-7.65.1.tar.xz -> sources/curl-7.65.1.tar.xz
downloading: sources/curl-7.65.1.tar.xz - 2.3MB of 2.3MB (100%)
building: curl-v7.65.1-arm-rtems5-1
sizes: curl-v7.65.1-arm-rtems5-1: 87.055MB (installed: 2.238MB)
cleaning: curl-v7.65.1-arm-rtems5-1
reporting: ftp/curl-7.65.1-1.cfg -> curl-v7.65.1-arm-rtems5-1.txt
reporting: ftp/curl-7.65.1-1.cfg -> curl-v7.65.1-arm-rtems5-1.xml
installing: curl-v7.65.1-arm-rtems5-1 -> /Users/chris/development/cs/rtems/5
cleaning: curl-v7.65.1-arm-rtems5-1
Build Set: Time 0:01:10.006872
13.4.3. Adding¶
Adding a package requires you first build it manually by downloading the source for the package and building it for RTEMS using the command line of a standard shell. If the package has not been ported to RTEMS you will need to port it and this may require asking questions on the package’s user or development support lists as well as RTEMS’s developers list. Your porting effort may end up with a patch. RTEMS requires a patch be submitted upstream to the project’s community as well as RTEMS. The RTEMS submission is best as a patch attached to ticket in Trac. A patch attached to a ticket can be referenced by an RSB configuration file and used in a build.
A package may create executables, for example Curl normally creates an executable called curl how ever it will probailty not run because the needed RTEMS configuration is not suitable. If found the RSB automatically adds the RTEMS library librtemsdefaultconfig.a to the LIBS variable used to link executables. This library provides a limited configuraiton suitable for linking an executable however it is not a set up that allows the resulting executable to run correctly. As a result it is best not to install these executables.
A custom RTEMS patch to an executate’s source code can turn it into a function that can be called by the RTEMS shell. Users can call the function in their executables simulating the running of the package’s command. If the package does not export the code in a suitable manner please contact the project’s community and see if you can work with them to provide a way for the code to be exported. This may be difficult because exporting internal headers and functions opens the project up to API compatibility issues they did not have before. In the simplest case attempting to get the code into a static library with a single call entry point exported in a header would give RTEMS user’s access to the package’s main functionality.
A package requires at least three (3) files to be created:
- Published Package Name:
The first file is the RTEMS build set file and it resides under the
rtems/configpath in a directory tree based on the FreeBSD ports collection. For the Curl package and RTEMS 5 this isrtems/config/ftp/curl.bset. If you do not know the FreeBSD port path for the package you are adding please ask. The build set file references a specific configuration file therefore linking the RTEMS version to a specific version of the package you are adding. Updating the package to a new version requires changing the build set to the new configuration file.- Package Version Configuration File:
The second file is an RTEMS version specific configuration file and it includes the RSB RTEMS BSP support. These configuration files reside in the
rtems/configtree and under the FreeBSD port’s path name. For example the Curl package is found in theftpdirectory of the FreeBSD ports tree so the Curl configuration path isrtems/config/ftp/curl-7.65.1-1.cfgfor that specific version. The configuration file name typically provides version specific references and the RTEMS build set file references a specific version. This configuration file references the build configuration file held in the common configuration file tree. An SHA512 hash is required to verify the source package that is downloaded.- Build Configuration File:
The build configuration. This is a common script that builds the package. It resides in the
source-builder/configdirectory and typically has the packages’s name with the major version number. If the build script does not change for each major version number a common base script can be created and included by each major version configuration script. The gcc compiler configuration is an example. This approach lets you branch a version if something changes that is not backwards compatible. It is important to keep existing versions building. The build configuration should be able to build a package for the build host as well as RTEMS as the RSB abstracts the RTEMS specific parts. See Configuration for more details.
13.4.4. Host and Build Flags¶
A package’s build is controlled by setting the compiler names and flags that are used when building. The RSB provides a macro called %{host_build_flags} to define these flags for you. Use this macro in the `%build section of your config script file to define the set up needed to build a native package or to cross-compile to a specific host such as RTEMS . The typical %build section is:
%build
build_top=$(pwd)
%{build_directory}
mkdir -p ${build_dir}
cd ${build_dir}
%{host_build_flags}
../${source_dir_curl}/configure \
--host=%{_host} \
--prefix=%{_prefix} \
--bindir=%{_bindir} \
--exec_prefix=%{_exec_prefix} \
--includedir=%{_includedir} \
--libdir=%{_libdir} \
--libexecdir=%{_libexecdir} \
--mandir=%{_mandir} \
--infodir=%{_infodir} \
--datadir=%{_datadir}
%{__make} %{?_smp_mflags} all
cd ${build_top}
The %{host_build_flags} checks if the build is native for the development host or a cross-compile build.
For a cross-complication build the flags are:
CC,CC_FOR_HOST:The C compiler used to build the package. For an RTEMS build this is the RTEMS C compiler. For example the ARM architecture and RTEMS 5 the value is set to
arm-rtems5-gcc.CXX,CXX_FOR_HOST:The C++ compiler used to build the package. For an RTEMS build this is the RTEMS C++ compiler. For example the ARM architecture and RTEMS 5 the value is set to
arm-rtems5-g++.CPPFLAGS,CPPFLAGS_FOR_HOST:The C compiler preprocessor flags used to build the package. Set any include paths in this variable as some configure scripts will warns you if include paths are set in the
CFLAGS.CFLAGS,CFLAGS_FOR_HOST:The C compiler flags used when running the C compiler. Set any include paths in the
CPPFLAGSvariable as some configure scripts will warn you if include paths in this variable.CXXFLAGS,CXXFLAGS_FOR_HOST:The C++ compiler flags used when running the C++ compiler. Set any include paths in the
CPPFLAGSvariable as some configure scripts will warn you if include paths in this variable.LDFLAGS,LDFLAGS_FOR_HOST:The linker flags used whne link package executables. The C or C++ compiler is used to run the linker.
LIBS,LIBS_FOR_HOST:A list of libraries passed to the linker when linking an executable.
CC_FOR_BUILD:The native C compiler.
CXX_FOR_BUILD:The native C++ compiler.
CPPFLAGS_FOR_BUILD:The C preprocessor flags used when preprocessoring a native C source file.
CFLAGS_FOR_BUILD:The native C compiler flags used when running the native C compiler.
CXXFLAGS_FOR_BUILD:The native C++ compiler flags used when running the native C++ compiler.
LDFLAGS_FOR_BUILD:The native linker flags used when linking a native executable.
LIBS_FOR_BUILD:The native libraries used to when linking a native executable.
For a native build the flags are:
CC,CC_FOR_BUILD:The native C compiler.
CXX,CXX_FOR_BUILD:The native C++ compiler.
CPPFLAGS,CPPFLAGS_FOR_BUILD:The C preprocessor flags used when preprocessoring a native C source file.
CFLAGS,CFLAGS_FOR_BUILD:The native C compiler flags used when running the native C compiler.
CXXFLAGS,CXXFLAGS_FOR_BUILD:The native C++ compiler flags used when running the native C++ compiler.
LDFLAGS,LDFLAGS_FOR_BUILD:The native linker flags used when linking a native executable.
LIBS,LIBS_FOR_BUILD:The native libraries used to when linking a native executable.
13.4.5. BSP Support¶
The RSB provides support to build packages for RTEMS. RTEMS applications can be viewed as statically linked executables operating in a single address space. As a result only the static libraries a package builds are required and these libraries need to be ABI compatible with the RTEMS kernel and application code. This means the compiler ABI flags used to build all the code in the executable must be the same. A 3rd party package must use the same compiler flags as the BSP used to build RTEMS.
Note
RTEMS’s dynamic loading support does not use the standard shared library support found in Unix and the ELF standard. RTEMS’s loader uses static libraries and the runtime link editor performs a similar function to a host based static linker. RTEMS will only reference static libraries even if dynamic libraries are created and installed.
The RSB provides the configuration file rtems/config/rtems-bsp.cfg to support building third-party packages and you need to include this file in your RTEMS version specific configuration file. For example the Curl configuration file rtems/config/curl/curl-7.65.1-1.cfg:
#
# Curl 7.65.1
#
%if %{release} == %{nil}
%define release 1 <1>
%endif
%include %{_configdir}/rtems-bsp.cfg <2>
#
# Curl Version
#
%define curl_version 7.65.1 <3>
%hash sha512 curl-%{curl_version}.tar.xz aba2d979a...72b6ac55df4 <4>
#
# Curl Build configuration
#
%include %{_configdir}/curl-1.cfg <5>
Items:
The release number.
Include the RSB RTEMS BSP support.
The Curl package’s version.
The SHA512 hash for the source file. The hash here has been shortened.
The Curl standard build configuration.
The RSB RTEMS BSP support file rtems/config/rtems-bsp.cfg checks to make sure the required RSB command line options are provided. These include --host and --with-rtems-bsp. If the --with-tools command line option is not given the ${_prefix} is used as the path to the tools. If the --with-rtems command line option is not given the ${_prefix} is used as the path to the installed RTEMS BSP.
Note
The RTEMS BSP and any dependent 3rd party packages must be installed to be seen as available. A path to the location the BSP has been built will not work.
The first check is to make sure a target is not specified. This is only used for Canadian cross-compilication builds and currently there is no support for RTEMS third party packages to build that way:
#
# The target is used by compilers or Cxc builds.
#
%if %{_target} != %{nil}
%error RTEMS BSP builds use --host and not --target
%endif
A host is required using the --host option:
#
# We need a host from the user to specifiy the RTEMS architecture and major
# version.
#
%if %{_host} == %{nil} && %{rtems_bsp_error} <1>
%error No RTEMS host or BSP specified: --host=<arch>-rtems<ver>
%endif
An RTEMS BSP is required using the --with-bsp option:
#
# We need a BSP from the user.
#
%ifn %{defined with_rtems_bsp}
%if %{rtems_bsp_error}
%error No RTEMS BSP specified: --rtems-bsp=arch/bsp (or --with-rtems-bsp=bsp)
%endif
%define with_rtems_bsp sparc/erc32
%endif
Check if the --with-tools or --with-rtems options have been provided and if they are not provided use the --prefix path:
#
# If no tools or RTEMS provided use the prefix.
#
%ifn %{defined with_tools}
%define with_tools %{_prefix}
%endif
%ifn %{defined with_rtems}
%define with_rtems %{_prefix}
%endif
Add the tools path to the envnironment path:
#
# Set the path to the tools.
#
%{path prepend %{with_tools}/bin}
RTEMS exports the build configuration in pkgconfig (.pc) files. The RSB can read these files even when there is no pkgconfig support installed on your development machine. The pkgconfig support provides a BSP’s configuration and the RSB uses it to set the followng RSB macros variables:
%{pkgconfig prefix %{_prefix}/lib/pkgconfig} <1>
%{pkgconfig crosscompile yes} <2>
%{pkgconfig filter-flags yes} <3>
#
# The RTEMS BSP Flags
#
%define rtems_bsp %{with_rtems_bsp}
%define rtems_bsp_ccflags %{pkgconfig ccflags %{_host}-%{rtems_bsp}} <4>
%define rtems_bsp_cflags %{pkgconfig cflags %{_host}-%{rtems_bsp}}
%define rtems_bsp_ldflags %{pkgconfig ldflags %{_host}-%{rtems_bsp}}
%define rtems_bsp_libs %{pkgconfig libs %{_host}-%{rtems_bsp}}
Items:
Set the path to the BSP’s pkgconfig file.
Let pkgconfig know this is a cross-compile build.
Filter flags such as warnings. Warning flags are specific to a package and RTEMS exports it’s warnings flags in the BSP configuration settings.
Ask pkgconfig for the various settings we require.
The flags obtained by pkgconfig and given a rtems_bsp prefix are used to set the RTEMS host variables CFLAGS, LDFLAGS and LIBS. When we build a third party library your host computer is the build machine and RTEMS is the host machine therefore we set the host variables:
%define host_cflags %{rtems_bsp_cflags}
%define host_ldflags %{rtems_bsp_ldflags}
%define host_libs %{rtems_bsp_libs}
Finally we provide all the paths you may require when configuring a package. Packages by default consider the _prefix the base and install various files under this tree. The package you are building is specific to a BSP and needs to install it’s files into the RTEMS specific BSP path under the _prefix. This allows more than BSP build of this package to be installed under the same _prefix at the same time:
%define rtems_bsp_prefix %{_prefix}/%{_host}/%{rtems_bsp} <1>
%define _exec_prefix %{rtems_bsp_prefix}
%define _bindir %{_exec_prefix}/bin
%define _sbindir %{_exec_prefix}/sbin
%define _libexecdir %{_exec_prefix}/libexec
%define _datarootdir %{_exec_prefix}/share
%define _datadir %{_datarootdir}
%define _sysconfdir %{_exec_prefix}/etc
%define _sharedstatedir %{_exec_prefix}/com
%define _localstatedir %{_exec_prefix}/var
%define _includedir %{_libdir}/include
%define _lib lib
%define _libdir %{_exec_prefix}/%{_lib}
%define _libexecdir %{_exec_prefix}/libexec
%define _mandir %{_datarootdir}/man
%define _infodir %{_datarootdir}/info
%define _localedir %{_datarootdir}/locale
%define _localedir %{_datadir}/locale
%define _localstatedir %{_exec_prefix}/var
Items:
The path to the installed BSP.
When you configure a package you can reference these paths and the RSB will provide sensible default or in this case map them to the BSP:
../${source_dir_curl}/configure \ <1>
--host=%{_host} \
--prefix=%{_prefix} \
--bindir=%{_bindir} \
--exec_prefix=%{_exec_prefix} \
--includedir=%{_includedir} \
--libdir=%{_libdir} \
--libexecdir=%{_libexecdir} \
--mandir=%{_mandir} \
--infodir=%{_infodir} \
--datadir=%{_datadir}
Items:
The configure command for Curl.
13.4.6. BSP Configuration¶
The following RSB macros are defined when building a package for RTEMS:
Note
A complete list can be obtained by building with the --trace flag. The log will contain a listing of all macros before and after the configuration is loaded.
%{rtems_bsp}:The name of the RTEMS BSP.
%{rtems_bsp_cc}:The C compiler name for the RTEMS BSP.
%{rtems_bsp_cflags}:The C compiler flags for the RTEMS BSP.
%{rtems_bsp_ccflags}:The C++ compiler flags for the RTEMS BSP.
%{rtems_bsp_incpath}:The include path to teh RTEMS BSP header files.
%{rtems_bsp_ldflags}:The linker flags for the RTEMS BSP.
%{rtems_bsp_libs}:The libraries used when linking an RTEMS BSP executable.
%{rtems_bsp_prefix}:The prefix for the RTEMS BSP.
%{rtems-libbsd}:The variable is set to
foundif LibBSD is available.%{rtems-defaultconfig}:The path of the RSB helper script to locate find header files or libraries.
%{_host}The host triplet passed on the command line to the set builder using the
--hostoptions. This is the RTEMS architecture and version. For examplearm-rtems5.%{host_cflags}:The BSP
CFLAGSreturned bypkgconfig.%{host_cxxflags}:The BSP
CXXFLAGSreturned bypkgconfig.%{host_includes}:The BSP include paths returned by
pkgconfig.%{host_ldflags}:The BSP
LDFLAGSreturned bypkgconfig.%{host_libs}:The libraries needed to be linked to create an executable. If LibBSD is installed the library
-lbsdis added. If the BSP has installed the RTEMS default configuration library (-lrtemsdefaultconfig) it is added to the list of libraries.%{host_build_flags}:This macro is defined in
defaults.mcand is a series of shell commands that set up the environment to build an RTEMS package. If the host and the build triplets are the same it is a native build for your development host. If the host is not the build machine it is a cross-complitation build. For either case the following are defined.%{_host_os}:The host operating system extracted from the
--hostcommand line option. For example the operating sstem for the host ofarm-rtems5isrtems5.%{_host_arch}:The host architecture extracted from the
--hostcommand line option. For example the architecture for the host ofarm-rtems5isarm.%{_host_cpu}:The host cpu extracted from the
--hostcommand line option. For example the cpu for the host ofarm-rtems5isarm.
13.5. Configuration¶
The RTEMS Source Builder has two types of configuration data:
Build Sets
Package Build Configurations
By default these files can be located in two separate directories and searched. The first directory is config in your current working directory (_topdir) and the second is config located in the base directory of the RTEMS Source Builder command you run (_sbdir). The RTEMS directory rtems` located at the top of the RTEMS Source Builder source code is an example of a specific build configuration directory. You can create custom or private build configurations and if you run the RTEMS Source Builder command from that directory your configurations will be used.
The configuration search path is a macro variable and is reference as %{_configdir}. It’s default is defined as:
_configdir : dir optional %{_topdir}/config:%{_sbdir}/config
Items:
The
_topdiris the directory you run the command from and_sbdiris the location of the RTEMS Source Builder command.A macro definition in a macro file has 4 fields, the label, type, constraint and the definition.
Build set files have the file extension .bset and the package build configuration files have the file extension of .cfg. The sb-set-builder command will search for build sets and the sb-builder commands works with package build configuration files.
Both types of configuration files use the # character as a comment character. Anything after this character on the line is ignored. There is no block comment.
13.5.1. Source and Patches¶
The RTEMS Source Builder provides a flexible way to manage source. Source and patches are declare in configurations file using the source and patch directives. These are a single line containing a Universal Resource Location or URL and can contain macros and shell expansions. The %prep section details the source and patch directives
The URL can reference remote and local source and patch resources. The following schemes are provided:
http:Remote access using the HTTP protocol.
https:Remote access using the Secure HTTP protocol.
ftp:Remote access using the FTP protocol.
git:Remote access to a GIT repository.
pm:Remote access to a patch management repository.
file:Local access to an existing source directory.
13.5.1.1. HTTP, HTTPS, and FTP¶
Remote access to TAR or ZIP files is provided using HTTP, HTTPS and FTP protocols. The full URL provided is used to access the remote file including any query components. The URL is parsed to extract the file component and the local source directory is checked for that file. If the file is located locally the remote file is not downloaded. Currently no other checks are made. If a download fails you need to manually remove the file from the source directory and start the build process again.
The URL can contain macros. These are expanded before issuing the request to download the file. The standard GNU GCC compiler source URL is:
%source set gcc ftp://ftp.gnu.org/gnu/gcc/gcc-%{gcc_version}/gcc-%{gcc_version}.tar.bz2
Items:
The
%sourcecommand’ssetcommand sets the source. The first is set and following sets are ignored.The source package is part of the
gccgroup.
The type of compression is automatically detected from the file extension. The supported compression formats are:
gz:GNU ZIP
bzip2:BZIP2
zip:ZIP
xz:XZ
The output of the decompression tool is fed to the standard tar utility if not a ZIP file and unpacked into the build directory. ZIP files are unpacked by the decompression tool and all other files must be in the tar file format.
The %source directive typically supports a single source file tar or zip file. The set command is used to set the URL for a specific source group. The first set command encountered is registered and any further set commands are ignored. This allows you to define a base standard source location and override it in build and architecture specific files. You can also add extra source files to a group. This is typically done when a collection of source is broken down in a number of smaller files and you require the full package. The source’s setup command must reside in the %prep: section and it unpacks the source code ready to be built.
If the source URL references the GitHub API server a tarball of the specified version is download. For example the URL for a Newlib snapshot GitHub is:
%define newlib_version 08eab6396f678cf5e5968acaed0bae9fd129983b
%define newlib_external 1
%define newlib_expand_name sourceware-mirror-newlib-cygwin-%{newlib_version}
%source set newlib --rsb-file=newlib-%{newlib_version}.tar.gz \
https://codeload.github.com/RTEMS/sourceware-mirror-newlib-cygwin/tar.gz/%{newlib_version}
13.5.1.2. GIT¶
A GIT repository can be cloned and used as source. The GIT repository resides in the ‘source’ directory under the git directory. You can edit, update and use the repository as you normally do and the results will used to build the tools. This allows you to prepare and test patches in the build environment the tools are built in. The GIT URL only supports the GIT protocol. You can control the repository via the URL by appending options and arguments to the GIT path. The options are delimited by ? and option arguments are delimited from the options with =. The options are:
protocol:Use a specific protocol. The supported values are
ssh,git,http,https,ftp,ftps,rsync, andnone.branch:Checkout the specified branch.
pull:Perform a pull to update the repository.
fetch:Perform a fetch to get any remote updates.
reset:Reset the repository. Useful to remove any local changes. You can pass the
hardargument to force a hard reset.
An example is:
%source set gcc git://gcc.gnu.org/git/gcc.git?branch=gcc-4_7-branch?reset=hard
This will clone the GCC git repository and checkout the 4.7-branch and perform a hard reset. You can select specific branches and apply patches. The repository is cleaned up before each build to avoid various version control errors that can arise.
The protocol option lets you set a specific protocol. The git:// prefix used by the RSB to select a git repository can be removed using none or replaced with one of the standard git protocols.
13.5.2. Macros and Defaults¶
The RTEMS Source Builder uses tables of macros read in when the tool runs. The initial global set of macros is called the defaults. These values are read from a file called defaults.mc and modified to suite your host. This host specific adaption lets the Source Builder handle differences in the build hosts.
Build set and configuration files can define new values updating and extending the global macro table. For example builds are given a release number. This is typically a single number at the end of the package name. For example:
%define release 1
Once defined if can be accessed in a build set or package configuration file with:
%{release}
The sb-defaults command lists the defaults for your host. I will not include the output of this command because of its size:
$ ../source-builder/sb-defaults
A nested build set is given a separate copy of the global macro maps. Changes in one change set are not seen in other build sets. That same happens with configuration files unless inline includes are used. Inline includes are seen as part of the same build set and configuration and changes are global to that build set and configuration.
13.5.2.1. Macro Maps and Files¶
Macros are read in from files when the tool starts. The default settings are read from the defaults macro file called defaults.mc located in the top level RTEMS Source Builder command directory. User macros can be read in at start up by using the --macros command line option.
The format for a macro in macro files is:
Name Type Attribute String
where Name is a case insensitive macro name, the Type field is:
none:Nothing, ignore.
dir:A directory path.
exe:An executable path.
triplet:A GNU style architecture, platform, operating system string.
the Attribute field is:
none:Nothing, ignore
required:The host check must find the executable or path.
optional:The host check generates a warning if not found.
override:Only valid outside of the
globalmap to indicate this macro overrides the same one in theglobalmap when the map containing it is selected.undefine:Only valid outside of the
globalmap to undefine the macro if it exists in theglobalmap when the map containing it is selected. Theglobalmap’s macro is not visible but still exists.
and the String field is a single or tripled multiline quoted string. The ‘String’ can contain references to other macros. Macro that loop are not currently detected and will cause the tool to lock up.
Maps are declared anywhere in the map using the map directive:
# Comments
[my-special-map] <1>
_host: none, override, 'abc-xyz'
multiline: none, override, '''First line,
second line,
and finally the last line'''
Items:
The map is set to
my-special-map.
Any macro definitions following a map declaration are placed in that map and the default map is global when loading a file. Maps are selected in configuration files by using the %select directive:
%select my-special-map
Selecting a map means all requests for a macro first check the selected map and if present return that value else the global map is used. Any new macros or changes update only the global map. This may change in future releases so please make sure you use the override attribute.
The macro files specified on the command line are looked for in the _configdir paths. See the definition of _configdir in Configuration for details. Included files need to add the %{_configdir} macro to the start of the file.
Macro map files can include other macro map files using the %include directive. The macro map to build binutils, gcc, newlib, gdb and RTEMS from version control heads is:
#
# Build all tool parts from version control head.
#
%include %{_configdir}/snapshots/binutils-head.mc
%include %{_configdir}/snapshots/gcc-head.mc
%include %{_configdir}/snapshots/newlib-head.mc
%include %{_configdir}/snapshots/gdb-head.mc
Items:
The file is
config/snapshots/binutils-gcc-newlib-gdb-head.mc.
The macro map defaults to global at the start of each included file and the map setting of the macro file including the other macro files does not change.
13.5.2.2. Personal Macros¶
When the tools start to run they will load personal macros. Personal macros are in the standard format for macros in a file. There are two places personal macros can be configured. The first is the environment variable RSB_MACROS. If present the macros from the file the environment variable points to are loaded. The second is a file called .rsb_macros in your home directory. You need to have the environment variable HOME defined for this work.
13.5.3. Configuration Reports¶
A configuration report detailing a build configuration is generated by the sb-set-builder command. The report can also be generated separately from the build process by invoking the sb-report command on the build set used for the build, for example:
cd rtems
../source-builder/sb-reports 5/rtems-sparc
less 5-rtems-sparc.txt
13.5.3.1. Report Mailing¶
Configuration reports from a build can be mailed to a specific email address for logging and monitoring. Mailing requires a number of parameters to function. These are:
To mail address
From mail address
SMTP host
The to mail address is taken from the macro %{_mail_tools_to} and the default is rtems-tooltestresults at rtems.org. You can override the default with a personal or user macro file or via the command line option --mail-to.
The from mail address is taken from:
GIT configuration
User
.mailrcfileCommand line
If you have configured an email and name in git it will be used used. If you do not a check is made for a .mailrc file. The environment variable MAILRC is used if present else your home directory is check. If found the file is scanned for the from setting:
set from="Foo Bar <foo@bar>"
You can also support a from address on the command line with the --mail-from option.
The SMTP host is taken from the macro %{_mail_smtp_host} and the default is localhost. You can override the default with a personal or user macro file or via the command line option --smtp-host.
13.5.4. Build Set Files¶
Build set files lets you list the packages in the build set you are defining and have a file extension of .bset. Build sets can define macro variables, inline include other files and reference other build set or package configuration files.
Defining macros is performed with the %define macro:
%define _target m32r-rtems4.11
Inline including another file with the %include macro continues processing with the specified file returning to carry on from just after the include point:
%include rtems-4.11-base.bset
This includes the RTEMS 4.11 base set of defines and checks. The configuration paths as defined by _configdir are scanned. The file extension is optional.
You reference build set or package configuration files by placing the file name on a single line:
tools/rtems-binutils-2.22-1
The _configdir path is scanned for tools/rtems-binutils-2.22-1.bset or tools/rtems-binutils-2.22-1.cfg. Build set files take precedent over package configuration files. If tools/rtems-binutils-2.22-1 is a build set a new instance of the build set processor is created and if the file is a package configuration the package is built with the package builder. This all happens once the build set file has finished being scanned.
13.5.5. Configuration Control¶
The RTEMS Souce Builder is designed to fit within most verification and validation processes. All of the RTEMS Source Builder is source code. The Python code is source and comes with a commercial friendly license. All configuration data is text and can be read or parsed with standard text based tools.
File naming provides configuration management. A specific version of a package is captured in a specific set of configuration files. The top level configuration file referenced in a build set or passed to the sb-builder command relates to a specific configuration of the package being built. For example the RTEMS configuration file rtems-gcc-4.7.2-newlib-2.0.0-1.cfg creates an RTEMS GCC and Newlib package where the GCC version is 4.7.2, the Newlib version is 2.0.0, plus any RTEMS specific patches that related to this version. The configuration defines the version numbers of the various parts that make up this package:
%define gcc_version 4.7.2
%define newlib_version 2.0.0
%define mpfr_version 3.0.1
%define mpc_version 0.8.2
%define gmp_version 5.0.5
The package build options, if there are any are also defined:
%define with_threads 1
%define with_plugin 0
%define with_iconv 1
The generic configuration may provide defaults in case options are not specified. The patches this specific version of the package requires can be included:
Patch0: gcc-4.7.2-rtems4.11-20121026.diff
Finally including the GCC 4.7 configuration script:
%include %{_configdir}/gcc-4.7-1.cfg
The gcc-4.7-1.cfg file is a generic script to build a GCC 4.7 compiler with Newlib. It is not specific to RTEMS. A bare no operating system tool set can be built with this file.
The -1 part of the file names is a revision. The GCC 4.7 script maybe revised to fix a problem and if this fix effects an existing script the file is copied and given a -2 revision number. Any dependent scripts referencing the earlier revision number will not be effected by the change. This locks down a specific configuration over time.
13.5.6. Personal Configurations¶
The RSB supports personal configurations. You can view the RTEMS support in the rtems directory as a private configuration tree that resides within the RSB source. There is also the bare set of configurations. You can create your own configurations away from the RSB source tree yet use all that the RSB provides.
To create a private configuration change to a suitable directory:
$ cd ~/work
$ mkdir test
$ cd test
$ mkdir config
and create a config directory. Here you can add a new configuration or build set file. The section ‘Adding New Configurations’ details how to add a new confguration.
13.5.7. New Configurations¶
This section describes how to add a new configuration to the RSB. We will add a configuration to build the Device Tree Compiler. The Device Tree Compiler or DTC is part of the Flattened Device Tree project and compiles Device Tree Source (DTS) files into Device Tree Blobs (DTB). DTB files can be loaded by operating systems and used to locate the various resources such as base addresses of devices or interrupt numbers allocated to devices. The Device Tree Compiler source code can be downloaded from http://www.jdl.com/software. The DTC is supported in the RSB and you can find the configuration files under the bare/config tree. I suggest you have a brief look over these files.
13.5.7.1. Layering by Including¶
Configurations can be layered using the %include directive. The user invokes the outer layers which include inner layers until all the required configuration is present and the package can be built. The outer layers can provide high level details such as the version and the release and the inner layers provide generic configuration details that do not change from one release to another. Macro variables are used to provide the specific configuration details.
13.5.7.2. Configuration File Numbering¶
Configuration files have a number at the end. This is a release number for that configuration and it gives us the ability to track a specific configuration for a specific version. For example lets say the developers of the DTC package change the build system from a single makefile to autoconf and automake between version 1.3.0 and version 1.4.0. The configuration file used to build the package would change have to change. If we did not number the configuration files the ability to build 1.1.0, 1.2.0 or 1.3.0 would be lost if we update a common configuration file to build an autoconf and automake version. For version 1.2.0 the same build script can be used so we can share the same configuration file between version 1.1.0 and version 1.2.0. An update to any previous release lets us still build the package.
13.5.7.3. Common Configuration Scripts¶
Common configuration scripts that are independent of version, platform and architecture are useful to everyone. These live in the Source Builder’s configuration directory. Currently there are scripts to build binutils, expat, DTC, GCC, GDB and libusb. These files contain the recipes to build these package without the specific details of the versions or patches being built. They expect to be wrapped by a configuration file that ties the package to a specific version and optionally specific patches.
13.5.7.4. DTC Example¶
We will be building the DTC for your host rather than a package for RTEMS. We will create a file called source-builder/config/dtc-1-1.cfg. This is a common script that can be used to build a specific version using a general recipe. The file name is dtc-1-1.cfg where the cfg extension indicates this is a configuration file. The first 1 says this is for the major release 1 of the package and the last 1 is the build configuration version.
The file starts with some comments that detail the configuration. If there is anything unusual about the configuration it is a good idea to add something in the comments here. The comments are followed by a check for the release. In this case if a release is not provided a default of 1 is used:
#
# DTC 1.x.x Version 1.
#
# This configuration file configure's, make's and install's DTC.
#
%if %{release} == %{nil}
%define release 1
%endif
The next section defines some information about the package. It does not effect the build and is used to annotate the reports. It is recommended this information is kept updated and accurate:
Name: dtc-%{dtc_version}-%{_host}-%{release}
Summary: Device Tree Compiler v%{dtc_version} for target %{_target} on host %{_host}
Version: %{dtc_version}
Release: %{release}
URL: https://www.devicetree.org/
BuildRoot: %{_tmppath}/%{name}-root-%(%{__id_u} -n)
The next section defines the source and any patches. In this case there is a single source package and it can be downloaded using the HTTP protocol. The RSB knows this is GZip’ped tar file. If more than one package is needed, add them increasing the index. The gcc-8-1.cfg configuration contains examples of more than one source package as well as conditionally including source packages based on the outer configuration options:
#
# Source
#
%source set dtc https://www.kernel.org/pub/software/utils/dtc/dtc-%{dtc_version}.tar.gz
The remainder of the script is broken in to the various phases of a build. They are:
. Preparation . Building . Installing, and . Cleaning
Preparation is the unpacking of the source, applying any patches as well as any package specific set ups. This part of the script is a standard Unix shell script. Be careful with the use of % and $. The RSB uses % while the shell scripts use $.
A standard pattern you will observe is the saving of the build’s top directory. This is used instead of changing into a subdirectory and then changing to the parent when finished. Some hosts will change in a subdirectory that is a link however changing to the parent does not change back to the parent of the link rather it changes to the parent of the target of the link and that is something the RSB nor you can track easily. The RSB configuration script’s are a collection of various subtle issues so please ask if you are unsure why something is being done a particular way.
The preparation phase will often include source and patch setup commands. Outer layers can set the source package and add patches as needed while being able to use a common recipe for the build. Users can override the standard build and supply a custom patch for testing using the user macro command line interface:
#
# Prepare the source code.
#
%prep
build_top=$(pwd)
%source setup dtc -q -n dtc-v%{dtc_version}
%patch setup dtc -p1
cd ${build_top}
The configuration file gcc-common-1.cfg is a complex example of source preparation. It contains a number of source packages and patches and it combines these into a single source tree for building. It uses links to map source into the GCC source tree so GCC can be built using the single source tree method. It also shows how to fetch source code from version control. Newlib is taken directly from its CVS repository.
Next is the building phase and for the DTC example this is simply a matter of running make. Note the use of the RSB macros for commands. In the case of %{__make} it maps to the correct make for your host. In the case of BSD systems we need to use the BSD make and not the GNU make.
If your package requires a configuration stage you need to run this before the make stage. Again the GCC common configuration file provides a detailed example:
%build
build_top=$(pwd)
cd dtc-v%{dtc_version}
%{build_build_flags}
%{__make} PREFIX=%{_prefix}
cd ${build_top}
You can invoke make with the macro %{?_smp_flags} as a command line argument. This macro is controlled by the --jobs command line option and the host CPU detection support in the RSB. If you are on a multicore host you can increase the build speed using this macro. It also lets you disabled building on multicores to aid debugging when testing.
Next is the install phase. This phase is a little more complex because you may be building a tar file and the end result of the build is never actually installed into the prefix on the build host and you may not even have permissions to perform a real install. Most packages install to the prefix and the prefix is typically supplied via the command to the RSB or the package’s default is used. The default can vary depending on the host’s operating system. To install to a path that is not the prefix the DESTDIR make variable is used. Most packages should honour the DESTDIR make variables and you can typically specify it on the command line to make when invoking the install target. This results in the package being installed to a location that is not the prefix but one you can control. The RSB provides a shell variable called SB_BUILD_ROOT you can use. In a build set where you are building a number of packages you can collect all the built packages in a single tree that is captured in the tar file.
Also note the use of the macro %{__rmdir}. The use of these macros allow the RSB to vary specific commands based on the host. This can help on hosts like Windows where bugs can effect the standard commands such as rm. There are many many macros to help you. You can find these listed in the defaults.mc file and in the trace output. If you are new to creating and editing configurations learning these can take a little time:
%install
build_top=$(pwd)
%{__rmdir} -rf $SB_BUILD_ROOT
cd dtc-v%{dtc_version}
%{__make} DESTDIR=$SB_BUILD_ROOT PREFIX=%{_prefix} install
cd ${build_top}
Finally there is an optional clean section. The RSB will run this section if --no-clean has not been provided on the command line. The RSB does clean up for you.
Once we have the configuration files we can execute the build using the sb-builder command. The command will perform the build and create a tar file in the tar directory:
$ ../source-builder/sb-builder --prefix=/usr/local \
--log=log_dtc devel/dtc-1.2.0
RTEMS Source Builder, Package Builder v0.2.0
config: devel/dtc-1.2.0
package: dtc-1.2.0-x86_64-freebsd9.1-1
download: https://www.kernel.org/pub/software/utils/dtc/dtc-v1.2.0.tgz -> sources/dtc-v1.2.0.tgz
building: dtc-1.2.0-x86_64-freebsd9.1-1
$ ls tar
dtc-1.2.0-x86_64-freebsd9.1-1.tar.bz2
If you want to have the package installed automatically you need to create a build set. A build set can build one or more packages from their configurations at once to create a single package. For example the GNU tools is typically seen as binutils, GCC and GDB and a build set will build each of these packages and create a single build set tar file or install the tools on the host into the prefix path.
The DTC build set file is called dtc.bset and contains:
#
# Build the DTC.
#
%define release 1
devel/dtc-1.2.0.cfg
To build this you can use something similar to:
$ ../source-builder/sb-set-builder --prefix=/usr/local --log=log_dtc \
--trace --bset-tar-file --no-install dtc
RTEMS Source Builder - Set Builder, v0.2.0
Build Set: dtc
config: devel/dtc-1.2.0.cfg
package: dtc-1.2.0-x86_64-freebsd9.1-1
building: dtc-1.2.0-x86_64-freebsd9.1-1
tarball: tar/x86_64-freebsd9.1-dtc-set.tar.bz2
cleaning: dtc-1.2.0-x86_64-freebsd9.1-1
Build Set: Time 0:00:02.865758
$ ls tar
dtc-1.2.0-x86_64-freebsd9.1-1.tar.bz2 x86_64-freebsd9.1-dtc-set.tar.bz2
The build is for a FreeBSD host and the prefix is for user installed packages. In this example I cannot let the source builder perform the install because I never run the RSB with root privileges so a build set or bset tar file is created. This can then be installed using root privileges.
The command also supplies the --trace option. The output in the log file will contain all the macros.
13.5.7.5. Debugging¶
New configuration files require debugging. There are two types of debugging. The first is debugging RSB script bugs. The --dry-run option is used here. Suppling this option will result in most of the RSB processing to be performed and suitable output placed in the log file. This with the --trace option should help you resolve any issues.
The second type of bug to fix are related to the execution of one of phases. These are usually a mix of shell script bugs or package set up or configuration bugs. Here you can use any normal shell script type debug technique such as set +x to output the commands or echo statements. Debugging package related issues may require you start a build with the RSB and supply --no-clean option and then locate the build directories and change directory into them and manually run commands until to figure what the package requires.
13.5.8. Scripting¶
Configuration files specify how to build a package. Configuration files are scripts and have a .cfg file extension. The script format is based loosely on the RPM spec file format however the use and purpose in this tool does not compare with the functionality and therefore the important features of the spec format RPM needs and uses.
The script language is implemented in terms of macros. The built-in list is:
%{}:Macro expansion with conditional logic.
%():Shell expansion.
%prep:The source preparation shell commands.
%build:The build shell commands.
%install:The package install shell commands.
%clean:The package clean shell commands.
%include:Inline include another configuration file.
%name:The name of the package.
%summary:A brief package description. Useful when reporting about a build.
%release:The package release. A number that is the release as built by this tool.
%version:The package’s version string.
%buildarch:The build architecture.
%source:Define a source code package. This macro has a number appended.
%patch:Define a patch. This macro has a number appended.
%hash:Define a checksum for a source or patch file.
%{echo message}:Print the following string as a message.
%{warning message}:Print the following string as a warning and continue.
%{error message}:Print the following string as an error and exit.
%select:Select the macro map. If there is no map nothing is reported.
%define:Define a macro. Macros cannot be redefined, you must first undefine it.
%undefine:Undefine a macro.
%if:Start a conditional logic block that ends with a
%endif.%ifn:Inverted start of a conditional logic block.
%ifarch:Test the architecture against the following string.
%ifnarch:Inverted test of the architecture
%ifos:Test the host operating system.
%else:Start the else conditional logic block.
%endfi:End the conditional logic block.
%bconf_with:Test the build condition with setting. This is the
--with-*command line option.%bconf_without:Test the build condition without setting. This is the
--without-*command line option.
13.5.8.1. Expanding¶
A macro can be %{string} or the equivalent of %string. The following macro expansions supported are:
%{string}:Expand the ‘string’ replacing the entire macro text with the text in the table for the entry ‘string . For example if ‘var’ is ‘foo’ then
${var}would becomefoo.%{expand: string}:Expand the ‘string’ and then use it as a
stringto the macro expanding the macro. For example iffoois set tobarandbaris set tofoobarthen%{expand:foo}would result infoobar. Shell expansion can also be used.%{with string}:Expand the macro to
1if the macrowith_stringis defined else expand to0. Macros with the namewith_stringcan be define with command line arguments to the RTEMS Source Builder commands.%{defined string}:Expand the macro to
1if a macro of namestringis defined else expand to ‘0’.%{?string: expression}:Expand the macro to
expressionif a macro of namestringis defined else expand to%{nil}.%{!?string: expression}:Expand the macro to
expressionif a macro of namestringis not defined. If the macro is define expand to%{nil}.%(expression):Expand the macro to the result of running the
expressionin a host shell. It is assumed this is a Unix type shell. For example%(whoami)will return your user name and%(date)will return the current date string.
13.5.8.2. %prep¶
The %prep macro starts a block that continues until the next block macro. The prep or preparation block defines the setup of the package’s source and is a mix of RTEMS Source Builder macros and shell scripting. The sequence is typically %source macros for source, %patch macros to patch the source mixed with some shell commands to correct any source issues:
%source setup gcc -q -c -T -n %{name}-%{version}
Items:
The source group to set up is
gcc.The source’s name is the macro
%{name}.The version of the source is the macro
%{version}.
The source set up are declared with the source set and add commands. For example:
%source set gdb http://ftp.gnu.org/gnu/gdb/gdb-%{gdb_version}.tar.bz2
This URL is the primary location of the GNU GDB source code and the RTEMS Source Builder can download the file from this location and by inspecting the file extension use bzip2 decompression with tar. When the %prep section is processed a check of the local source directory is made to see if the file has already been downloaded. If not found in the source cache directory the package is downloaded from the URL. You can append other base URLs via the command line option --url. This option accepts a comma delimited list of sites to try.
You could optionally have a few source files that make up the package. For example GNU’s GCC was a few tar files for a while and it is now a single tar file. Support for multiple source files can be conditionally implemented with the following scripting:
%source set gcc ftp://ftp.gnu.org/gnu/gcc/gcc-%{gcc_version}/gcc-code-%{gcc_version}.tar.bz2
%source add gcc ftp://ftp.gnu.org/gnu/gcc/gcc-%{gcc_version}/gcc-g++-%{gcc_version}.tar.bz2
%source setup gcc -q -T -D -n gcc-%{gcc_version}
Separate modules use separate source groups. The GNU GCC compiler for RTEMS uses Newlib, MPFR, MPC, and GMP source packages. You define the source with:
%source set gcc ftp://ftp.gnu.org/gnu/gcc/gcc-%{gcc_version}/gcc-%{gcc_version}.tar.bz2
%source set newlib ftp://sourceware.org/pub/newlib/newlib-%{newlib_version}.tar.gz
%source set mpfr http://www.mpfr.org/mpfr-%{mpfr_version}/mpfr-%{mpfr_version}.tar.bz2
%source set mpc http://www.multiprecision.org/mpc/download/mpc-%{mpc_version}.tar.gz
%source set gmp ftp://ftp.gnu.org/gnu/gmp/gmp-%{gmp_version}.tar.bz2
and set up with:
%source setup gcc -q -n gcc-%{gcc_version}
%source setup newlib -q -D -n newlib-%{newlib_version}
%source setup mpfr -q -D -n mpfr-%{mpfr_version}
%source setup mpc -q -D -n mpc-%{mpc_version}
%source setup gmp -q -D -n gmp-%{gmp_version}
Patching also occurs during the preparation stage. Patches are handled in a similar way to the source packages except you only add patches. Patches are applied using the setup command. The setup command takes the default patch option. You can provide options with each patch by adding them as arguments before the patch URL. Patches with no options uses the setup default.
%patch add gdb %{rtems_gdb_patches}/gdb-sim-arange-inline.diff
%patch add gdb -p0 %{rtems_gdb_patches}/gdb-sim-cgen-inline.diff
Items:
This patch has the custom option of
-p0.
To apply these patches:
%patch setup gdb -p1
Items:
The default options for
gdbset up.
13.5.8.3. %build¶
The %build macro starts a block that continues until the next block macro. The build block is a series of shell commands that execute to build the package. It assumes all source code has been unpacked, patch and adjusted so the build will succeed.
The following is an example take from the GitHub STLink project. The STLink is a JTAG debugging device for the ST ARM family of processors:
%build
export PATH="%{_bindir}:${PATH}"
cd texane-stlink-%{stlink_version}
./autogen.sh
%if "%{_build}" != "%{_host}"
CFLAGS_FOR_BUILD="-g -O2 -Wall" \
%endif
CPPFLAGS="-I $SB_TMPPREFIX/include/libusb-1.0" \
CFLAGS="$SB_OPT_FLAGS" \
LDFLAGS="-L $SB_TMPPREFIX/lib" \
./configure \
--build=%{_build} --host=%{_host} \
--verbose \
--prefix=%{_prefix} --bindir=%{_bindir} \
--exec-prefix=%{_exec_prefix} \
--includedir=%{_includedir} --libdir=%{_libdir} \
--mandir=%{_mandir} --infodir=%{_infodir}
%{__make} %{?_smp_mflags} all
cd ..
Items:
Set up the PATH environment variable by setting the
PATHenvironment variable. This is not always needed.This package builds in the source tree
texane-stlink-%{stlink_version}so enter it before building.The package is actually checked directly out from the github project and so it needs its
autoconfandautomakefiles generated. Invoke the provided scriptautogen.shIf the build machine and host are not the same the build is a cross-compile. Update the flags for a cross-compiled build.
The flags set in the environment before
configureare various settings that need to be passed to customise the build. In this example an include path is being set to the install point oflibusb. This package requireslibusbis built before it.The
configurecommand. The RTEMS Source Builder provides all the needed paths as macro variables. You just need to provide them toconfigure.Run
make. Do not usemakedirectly, use the RTEMS Source Builder’s defined value. This value is specific to the host. A large number of packages need GNU make and on BSD systems this isgmake. You can optionally add the SMP flags if the packages build system can handle parallel building with multiple jobs. The_smp_mflagsvalue is automatically setup for SMP hosts to match the number of cores the host has.
13.5.8.4. %install¶
The %install macro starts a block that continues until the next block macro. The install block is a series of shell commands that execute to install the package. You can assume the package has built correctly when this block starts executing.
Never install the package to the actual prefix the package was built with. Always install to the RTEMS Source Builder’s temporary path defined in the macro variable __tmpdir. The RTEMS Source Builder sets up a shell environment variable called SB_BUILD_ROOT as the standard install point. Most packages support adding DESTDIR= to the make install command.
Looking at the same example as in %build:
%install
export PATH="%{_bindir}:${PATH}" <1>
rm -rf $SB_BUILD_ROOT <2>
cd texane-stlink-%{stlink_version} <3>
%{__make} DESTDIR=$SB_BUILD_ROOT install <4>
cd ..
Items:
Setup the PATH environment variable. This is not always needed.
Clean any installed files. This makes sure the install is just what the package installs and not any left over files from a broken build or install.
Enter the build directory. In this example it just happens to be the source directory.
Run
make installto install the package overriding theDESTDIRmake variable.
13.5.8.5. %clean¶
The %clean macro starts a block that continues until the next block macro. The clean block is a series of shell commands that execute to clean up after a package has been built and install. This macro is currenly not been used because the RTEMS Source Builder automatically cleans up.
13.5.8.6. %include¶
The %include macro inline includes the specific file. The __confdir path is searched. Any relative path component of the include file is appended to each part of the __configdir. Adding an extension is optional as files with .bset and .cfg are automatically searched for.
Inline including means the file is processed as part of the configuration at the point it is included. Parsing continues from the next line in the configuration file that contains the %include macro.
Including files allow a kind of configuration file reuse. The outer configuration files provide specific information such as package version numbers and patches and then include a generic configuration script which builds the package:
%include %{_configdir}/gcc-4.7-1.cfg
13.5.8.7. %name¶
The name of the package being built. The name typically contains the components of the package and their version number plus a revision number. For the GCC with Newlib configuration the name is typically:
Name: %{_target}-gcc-%{gcc_version}-newlib-%{newlib_version}-%{release}
13.5.8.8. %summary¶
The %summary is a brief description of the package. It is useful when reporting. This information is not capture in the package anywhere. For the GCC with Newlib configuration the summary is typically:
Summary: GCC v%{gcc_version} and Newlib v%{newlib_version} for target %{_target} on host %{_host}
13.5.8.9. %release¶
The %release is a packaging number that allows revisions of a package to happen where no package versions change. This value typically increases when the configuration building the package changes:
%define release 1
13.5.8.10. %version¶
The %version macro sets the version the package. If the package is a single component it tracks that component’s version number. For example in the libusb configuration the %version is the same as %libusb_version, however in a GCC with Newlib configuration there is no single version number. In this case the GCC version is used:
Version: %{gcc_version}
13.5.8.11. %buildarch¶
The %buildarch macro is set to the architecture the package contains. This is currently not used in the RTEMS Source Builder and may go away. This macro is more important in a real packaging system where the package could end up on the wrong architecture.
13.5.8.12. %source¶
The %source macro has 3 commands that controls what it does. You can set the source files, add source files to a source group, and setup the source file group getting it ready to be used.
Source files are source code files in tar or zip files that are unpacked, copied or symbolically linked into the package’s build tree. Building a package requires one or more dependent packages. These are typically the packages source code plus dependent libraries or modules. You can create any number of these source groups and set each of them up with a separate source group for each needed library or module. Each source group normally has a single tar, zip or repository and the set defines this. Some projects split the source code into separate tar or zip files and you install them by using the add command.
The first instance of a set command creates the source group and sets the source files to be set up. Subsequent set commands for the same source group are ignored. this lets you define the standard source files and override them for specific releases or snapshots. To set a source file group:
%source set gcc ftp://ftp.gnu.org/gnu/gcc/gcc-%{gcc_version}/gcc-%{gcc_version}.tar.bz2
Items:
The source group is
gcc.
To add another source package to be installed into the same source tree you use the add command:
%source add gcc ftp://ftp.gnu.org/gnu/gcc/gcc-%{gcc_version}/g++-%{gcc_version}.tar.bz2
The source setup command can only be issued in the %prep: section. The setup is:
%source gcc setup -q -T -D -n %{name}-%{version}
Accepted options are:
-n:The
-noption is used to set the name of the software’s build directory. This is necessary only when the source archive unpacks into a directory named other than<name>-<version>.-c:The
-coption is used to direct%setupto create the top-level build directory before unpacking the sources.-D:The
-Doption is used to direct%setupto not delete the build directory prior to unpacking the sources. This option is used when more than one source archive is to be unpacked into the build directory, normally with the-bor-aoptions.-T:The
-Toption is used to direct %setup to not perform the default unpacking of the source archive specified by the firstSource:macro. It is used with the-aor-boptions.-b <n>:The
-boption is used to direct%setupto unpack the source archive specified on the nthSource:macro line before changing directory into the build directory.
13.5.8.13. %patch¶
The %patch macro has the same 3 command as the %source command however the set commands is not really that useful with the %patch command. You add patches with the add command and setup applies the patches. Patch options can be added to each patch by placing them before the patch URL. If no patch option is provided the default options passed to the setup command are used. An option starts with a -. The setup command must reside inside the %prep section.
Patches are grouped in a similar way to the %source macro so you can control applying a group of patches to a specific source tree.
The __patchdir path is searched.
To add a patch:
%patch add gcc gcc-4.7.2-rtems4.11-20121026.diff
%patch add gcc -p0 gcc-4.7.2-rtems4.11-20121101.diff
Items:
The patch group is
gcc.Option
-p0is this specific to this patch.
Placing %patch setup in the %prep section will apply the groups patches:
.. code-block:: spec
%patch setup gcc -p1
The patch group is
gcc.The default option used to apply the patch is
-p1.
13.5.8.14. %hash¶
The %hash macro requires 3 arguments and defines a checksum for a specific file. The checksum is not applied until the file is checked before downloading and once downloaded. A patch or source file that does not have a hash defined generates a warning.
A file to be checksummed must be unique in the source and patch directories. The basename of the file is used as the key for the hash.
The hash algorthim can be md5, sha1, sha224, sha256, sha384, and sha512 and we typically use md5.
To add a hash:
%hash md5 <1> net-snmp-%{net_snmp_version}.tar.gz <2> 7db683faba037249837b226f64d566d4 <3>
Items:
The type of checksum.
The file to checksum. It can contain macros that are expanded for you.
The MD5 hash for the Net-SNMP file
net-snmp-5.7.2.1.tar.gz.
Do not include a path with the file name. Only the basename is required. Files can be searched for from a number of places and having a path conponent would create confusion. This does mean files with hashes must be unique.
Downloading off repositories such as git and cvs cannot be checksummed. It is assumed those protocols and tools manage the state of the files.
13.5.8.15. %echo¶
The %echo macro outputs the following string to stdout. This can also be used as %{echo: message}.
13.5.8.16. %warning¶
The %warning macro outputs the following string as a warning. This can also be used as %{warning: message}.
13.5.8.17. %error¶
The %error macro outputs the follow string as an error and exits the RTEMS Source Builder. This can also be used as %{error: message}.
13.5.8.18. %select¶
The %select macro selects the map specified. If there is no map no error or warning is generated. Macro maps provide a simple way for a user to override the settings in a configuration file without having to edit it. The changes are recorded in the build report so they can be traced.
Configurations use different maps so macro overrides can target a specific package.
The default map is global:
%select gcc-4.8-snapshot <1>
%define one_plus_one 2 <2>
Items:
The map switches to
gcc-4.8-snapshot. Any overrides in this map will be used.Defining macros only updates the
globalmap and not the selected map.
13.5.8.19. %define¶
The %define macro defines a new macro or updates an existing one. If no value is given it is assumed to be 1:
%define foo bar
%define one_plus_one 2
%define one <1>
Items:
The macro _one_ is set to 1.
13.5.8.20. %undefine¶
The %undefine macro removes a macro if it exists. Any further references to it will result in an undefine macro error.
13.5.8.21. %if¶
The %if macro starts a conditional logic block that can optionally have a else section. A test follows this macro and can have the following operators:
%{} | Check the macro is set or true, ie non-zero: %if ${foo}
%warning The test passes, must not be empty or is non-zero
%else
%error The test fails, must be empty or zero
%endif
|
! | The not operator inverts the test of the macro: %if ! ${foo}
%warning The test passes, must be empty or zero
%else
%error The test fails, must not be empty or is non-zero
%endif
|
== | The left hand size must equal the right hand side. For example: %define one 1
%if ${one} == 1
%warning The test passes
%else
%error The test fails
%endif
You can also check to see if a macro is empty: %if ${nothing} == %{nil}
%warning The test passes
%else
%error The test fails
|
!= | The left hand size does not equal the right hand side. For example: #
# Check a value not being equal.
#
%define one 1
%if ${one} != 2
%warning The test passes
%else
%error The test fails
%endif
#
# Check if a macro is set.
#
%if ${something} != %{nil}
%warning The test passes
%else
%error The test fails
%endif
|
> | The left hand side is numerically greater than the right hand side. |
> | The left hand side is numerically greater than or equal to the right hand side. |
< | The left hand side is numerically less than the right hand side. |
<= | The left hand side is numerically less than or equal to the right hand side. |
13.5.8.22. %ifn¶
The %ifn macro inverts the normal %if logic. It avoids needing to provide empty if blocks followed by else blocks. It is useful when checking if a macro is defined:
%ifn %{defined foo}
%define foo bar
%endif
13.5.8.23. %ifarch¶
The %ifarch is a short cut for %if %{_arch} == i386. Currently not used.
13.5.8.24. %ifnarch¶
The %ifnarch is a short cut for %if %{_arch} != i386. Currently not used.
13.5.8.25. %ifos¶
The %ifos is a short cut for %if %{_os} != mingw32. It allows conditional support for various operating system differences when building packages.
13.5.8.26. %else¶
The %else macro starts the conditional else block.
13.5.8.27. %endfi¶
The %endif macro ends a conditional logic block.
13.5.8.28. %bconf_with¶
The %bconf_with macro provides a way to test if the user has passed a specific option on the command line with the --with-<label> option. This option is only available with the sb-builder command.
13.5.8.29. %bconf_without¶
The %bconf_without macro provides a way to test if the user has passed a specific option on the command line with the --without-<label> option. This option is only available with the sb-builder command.
13.6. Commands¶
13.6.1. Checker (sb-check)¶
This commands checks your system is set up correctly. Most options are ignored:
$ ../source-builder/sb-check --help
sb-check: [options] [args]
RTEMS Source Builder, an RTEMS Tools Project (c) 2012-2013 Chris Johns
Options and arguments:
--force : Force the build to proceed
--quiet : Quiet output (not used)
--trace : Trace the execution
--dry-run : Do everything but actually run the build
--warn-all : Generate warnings
--no-clean : Do not clean up the build tree
--always-clean : Always clean the build tree, even with an error
--jobs : Run with specified number of jobs, default: num CPUs.
--host : Set the host triplet
--build : Set the build triplet
--target : Set the target triplet
--prefix path : Tools build prefix, ie where they are installed
--topdir path : Top of the build tree, default is $PWD
--configdir path : Path to the configuration directory, default: ./config
--builddir path : Path to the build directory, default: ./build
--sourcedir path : Path to the source directory, default: ./source
--patchdir path : Path to the patches directory, default: ./patches
--tmppath path : Path to the temp directory, default: ./tmp
--macros file[,[file] : Macro format files to load after the defaults
--log file : Log file where all build out is written too
--url url[,url] : URL to look for source
--no-download : Disable the source downloader
--targetcflags flags : List of C flags for the target code
--targetcxxflags flags : List of C++ flags for the target code
--libstdcxxflags flags : List of C++ flags to build the target libstdc++ code
--with-<label> : Add the --with-<label> to the build
--without-<label> : Add the --without-<label> to the build
--regression : Set --no-install, --keep-going and --always-clean
$ ../source-builder/sb-check
RTEMS Source Builder - Check, v0.2.0
Environment is ok
13.6.2. Defaults (sb-defaults)¶
This commands outputs and the default macros for your when given no arguments. Most options are ignored:
$ ../source-builder/sb-defaults --help
sb-defaults: [options] [args]
RTEMS Source Builder, an RTEMS Tools Project (c) 2012-2013 Chris Johns
Options and arguments:
--force : Force the build to proceed
--quiet : Quiet output (not used)
--trace : Trace the execution
--dry-run : Do everything but actually run the build
--warn-all : Generate warnings
--no-clean : Do not clean up the build tree
--always-clean : Always clean the build tree, even with an error
--jobs : Run with specified number of jobs, default: num CPUs.
--host : Set the host triplet
--build : Set the build triplet
--target : Set the target triplet
--prefix path : Tools build prefix, ie where they are installed
--topdir path : Top of the build tree, default is $PWD
--configdir path : Path to the configuration directory, default: ./config
--builddir path : Path to the build directory, default: ./build
--sourcedir path : Path to the source directory, default: ./source
--patchdir path : Path to the patches directory, default: ./patches
--tmppath path : Path to the temp directory, default: ./tmp
--macros file[,[file] : Macro format files to load after the defaults
--log file : Log file where all build out is written too
--url url[,url] : URL to look for source
--no-download : Disable the source downloader
--targetcflags flags : List of C flags for the target code
--targetcxxflags flags : List of C++ flags for the target code
--libstdcxxflags flags : List of C++ flags to build the target libstdc++ code
--with-<label> : Add the --with-<label> to the build
--without-<label> : Add the --without-<label> to the build
--regression : Set --no-install, --keep-going and --always-clean
13.6.3. Set Builder (sb-set-builder)¶
This command builds a set:
$ ../source-builder/sb-set-builder --help
RTEMS Source Builder, an RTEMS Tools Project (c) 2012-2013 Chris Johns
Options and arguments:
--force : Force the build to proceed
--quiet : Quiet output (not used)
--trace : Trace the execution
--dry-run : Do everything but actually run the build
--warn-all : Generate warnings
--no-clean : Do not clean up the build tree
--always-clean : Always clean the build tree, even with an error
--regression : Set --no-install, --keep-going and --always-clean
---jobs : Run with specified number of jobs, default: num CPUs.
--host : Set the host triplet
--build : Set the build triplet
--target : Set the target triplet
--prefix path : Tools build prefix, ie where they are installed
--topdir path : Top of the build tree, default is $PWD
--configdir path : Path to the configuration directory, default: ./config
--builddir path : Path to the build directory, default: ./build
--sourcedir path : Path to the source directory, default: ./source
--patchdir path : Path to the patches directory, default: ./patches
--tmppath path : Path to the temp directory, default: ./tmp
--macros file[,[file] : Macro format files to load after the defaults
--log file : Log file where all build out is written too
--url url[,url] : URL to look for source
--no-download : Disable the source downloader
--no-install : Do not install the packages to the prefix
--targetcflags flags : List of C flags for the target code
--targetcxxflags flags : List of C++ flags for the target code
--libstdcxxflags flags : List of C++ flags to build the target libstdc++ code
--with-<label> : Add the --with-<label> to the build
--without-<label> : Add the --without-<label> to the build
--mail-from : Email address the report is from.
--mail-to : Email address to send the email too.
--mail : Send email report or results.
--smtp-host : SMTP host to send via.
--no-report : Do not create a package report.
--report-format : The report format (text, html, asciidoc).
--bset-tar-file : Create a build set tar file
--pkg-tar-files : Create package tar files
--list-bsets : List available build sets
--list-configs : List available configurations
--list-deps : List the dependent files.
The arguments are a list of build sets to build.
Options:
--force:Force the build to proceed even if the host check fails. Typically this happens if executable files are found in the path at a different location to the host defaults.
--trace:Trace enable printing of debug information to stdout. It is really only of use to RTEMS Source Builder’s developers.
--dry-run:Do everything but actually run the build commands. This is useful when checking a new configuration parses cleanly.
--warn-all:Generate warnings.
--no-clean:Do not clean up the build tree during the cleaning phase of the build. This leaves the source and the build output on disk so you can make changes, or amend or generate new patches. It also allows you to review configure type output such as
config.log.--always-clean:Clean away the results of a build even if the build fails. This is normally used with
--keep-goingwhen regression testing to see which build sets fail to build. It keeps the disk usage down.--jobs:Control the number of jobs make is given. The jobs can be
nonefor only 1 job,halfso the number of jobs is half the number of detected cores, a fraction such as0.25so the number of jobs is a quarter of the number of detected cores and a number such as25which forces the number of jobs to that number.--host:Set the host triplet value. Be careful with this option.
--build:Set the build triplet. Be careful with this option.
--target:Set the target triplet. Be careful with this option. This is useful if you have a generic configuration script that can work for a range of architectures.
--prefix path:Tools build prefix, ie where they are installed.
--topdir path:Top of the build tree, that is the current directory you are in.
--configdir path:Path to the configuration directory. This overrides the built in defaults.
--builddir path:Path to the build directory. This overrides the default of build.
--sourcedir path:Path to the source directory. This overrides the default of source.
--patchdir path:Path to the patches directory. This overrides the default of patches.
--tmppath path:Path to the temporary directory. This overrides the default of tmp.
--macros files:Macro files to load. The configuration directory path is searched.
--log file:Log all the output from the build process. The output is directed to stdout if no log file is provided.
--url url:URL to look for source when downloading. This is can be comma separate list.
--no-download:Disable downloading of source and patches. If the source is not found an error is raised.
--targetcflags flags:List of C flags for the target code. This allows for specific local customisation when testing new variations.
--targetcxxflags flags:List of C++ flags for the target code. This allows for specific local customisation when testing new variations.
--libstdcxxflags flags:List of C++ flags to build the target libstdc++ code. This allows for specific local customisation when testing new variations.
--with-<label>:Add the
--with-<label>to the build. This can be tested for in a script with the%bconf_withmacro.--without-<label>:Add the
--without-<label>to the build. This can be tested for in a script with the%bconf_withoutmacro.--mail-from:Set the from mail address if report mailing is enabled.
--mail-to:Set the to mail address if report mailing is enabled. The report is mailed to this address.
--mail:Mail the build report to the mail to address.
--smtp-host:The SMTP host to use to send the email. The default is
localhost.--no-report:Do not create a report format.
--report-format format:The report format can be
textorhtml. The default ishtml.--keep-going:Do not stop on error. This is useful if your build sets performs a large number of testing related builds and there are errors.
--always-clean:Always clean the build tree even with a failure.
--no-install:Do not install the packages to the prefix. Use this if you are only after the tar files.
--regression:A convenience option which is the same as
--no-install,--keep-goingand--always-clean.--bset-tar-file:Create a build set tar file. This is a single tar file of all the packages in the build set.
--pkg-tar-files:Create package tar files. A tar file will be created for each package built in a build set.
--list-bsets:List available build sets.
--list-configs:List available configurations.
--list-deps:Print a list of dependent files used by a build set. Dependent files have a
dep[?]` prefix where ``?is a number. The files are listed alphabetically.
13.6.4. Set Builder (sb-builder)¶
This command builds a configuration as described in a configuration file. Configuration files have the extension of .cfg:
$ ./source-builder/sb-builder --help
sb-builder: [options] [args]
RTEMS Source Builder, an RTEMS Tools Project (c) 2012 Chris Johns
Options and arguments:
--force : Force the build to proceed
--quiet : Quiet output (not used)
--trace : Trace the execution
--dry-run : Do everything but actually run the build
--warn-all : Generate warnings
--no-clean : Do not clean up the build tree
--always-clean : Always clean the build tree, even with an error
--jobs : Run with specified number of jobs, default: num CPUs.
--host : Set the host triplet
--build : Set the build triplet
--target : Set the target triplet
--prefix path : Tools build prefix, ie where they are installed
--topdir path : Top of the build tree, default is $PWD
--configdir path : Path to the configuration directory, default: ./config
--builddir path : Path to the build directory, default: ./build
--sourcedir path : Path to the source directory, default: ./source
--patchdir path : Path to the patches directory, default: ./patches
--tmppath path : Path to the temp directory, default: ./tmp
--macros file[,[file] : Macro format files to load after the defaults
--log file : Log file where all build out is written too
--url url[,url] : URL to look for source
--targetcflags flags : List of C flags for the target code
--targetcxxflags flags : List of C++ flags for the target code
--libstdcxxflags flags : List of C++ flags to build the target libstdc++ code
--with-<label> : Add the --with-<label> to the build
--without-<label> : Add the --without-<label> to the build
--list-configs : List available configurations
13.7. Building and Deploying Tool Binaries¶
If you wish to create and distribute your build or you want to archive a build you can create a tar file. We term this deploying a build. This is a more advanced method for binary packaging and installing of tools.
By default the RTEMS Source Builder installs the built packages directly and optionally it can also create a build set tar file or a package tar file per package built. The normal and default behaviour is to let the RTEMS Source Builder install the tools. The source will be downloaded, built, installed and cleaned up.
The tar files are created with the full build prefix present and if you follow the examples given in this documentation the path is absolute. This can cause problems if you are installing on a host you do not have super user or administrator rights on because the prefix path may references part you do not have write access too and tar will not extract the files. You can use the --strip-components option in tar if your host tar application supports it to remove the parts you do not have write access too or you may need to unpack the tar file somewhere and copy the file tree from the level you have write access from. Embedding the full prefix path in the tar files lets you know what the prefix is and is recommended. For example if /home/chris/development/rtems/4.11 is the prefix used you cannot change directory to the root (/) and untar the file because the /home is root access only. To install a tar file you have downloaded into your new machine’s Downloads directory in your home directoty you would enter:
$ cd /somewhere
$ tar --strip-components=3 -xjf \
$HOME/Downloads/rtems-4.11-sparc-rtems4.11-1.tar.bz2
A build set tar file is created by adding --bset-tar-file option to the sb-set-builder command:
$ ../source-builder/sb-set-builder --log=l-sparc.txt \
--prefix=$HOME/development/rtems/4.11 \
--bset-tar-file \ <1>
4.11/rtems-sparc
Source Builder - Set Builder, v0.2.0
Build Set: 4.11/rtems-sparc
config: expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-freebsd9.1-1
building: expat-2.1.0-x86_64-freebsd9.1-1
config: tools/rtems-binutils-2.22-1.cfg
package: sparc-rtems4.11-binutils-2.22-1
building: sparc-rtems4.11-binutils-2.22-1
config: tools/rtems-gcc-4.7.2-newlib-1.20.0-1.cfg
package: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
building: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
config: tools/rtems-gdb-7.5.1-1.cfg
package: sparc-rtems4.11-gdb-7.5.1-1
building: sparc-rtems4.11-gdb-7.5.1-1
installing: rtems-4.11-sparc-rtems4.11-1 -> /home/chris/development/rtems/4.11 <2>
installing: rtems-4.11-sparc-rtems4.11-1 -> /home/chris/development/rtems/4.11
installing: rtems-4.11-sparc-rtems4.11-1 -> /home/chris/development/rtems/4.11
installing: rtems-4.11-sparc-rtems4.11-1 -> /home/chris/development/rtems/4.11
tarball: tar/rtems-4.11-sparc-rtems4.11-1.tar.bz2 <3>
cleaning: expat-2.1.0-x86_64-freebsd9.1-1
cleaning: sparc-rtems4.11-binutils-2.22-1
cleaning: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
cleaning: sparc-rtems4.11-gdb-7.5.1-1
Build Set: Time 0:15:25.92873
Items
The option to create a build set tar file.
The installation still happens unless you specify
--no-install.Creating the build set tar file.
You can also suppress installing the files using the --no-install option. This is useful if your prefix is not accessible, for example when building Canadian cross compiled tool sets:
$ ../source-builder/sb-set-builder --log=l-sparc.txt \
--prefix=$HOME/development/rtems/4.11 \
--bset-tar-file \
--no-install \ <1>
4.11/rtems-sparc
Source Builder - Set Builder, v0.2.0
Build Set: 4.11/rtems-sparc
config: expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-freebsd9.1-1
building: expat-2.1.0-x86_64-freebsd9.1-1
config: tools/rtems-binutils-2.22-1.cfg
package: sparc-rtems4.11-binutils-2.22-1
building: sparc-rtems4.11-binutils-2.22-1
config: tools/rtems-gcc-4.7.2-newlib-1.20.0-1.cfg
package: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
building: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
config: tools/rtems-gdb-7.5.1-1.cfg
package: sparc-rtems4.11-gdb-7.5.1-1
building: sparc-rtems4.11-gdb-7.5.1-1
tarball: tar/rtems-4.11-sparc-rtems4.11-1.tar.bz2 <2>
cleaning: expat-2.1.0-x86_64-freebsd9.1-1
cleaning: sparc-rtems4.11-binutils-2.22-1
cleaning: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
cleaning: sparc-rtems4.11-gdb-7.5.1-1
Build Set: Time 0:14:11.721274
$ ls tar
rtems-4.11-sparc-rtems4.11-1.tar.bz2
Items
The option to supressing installing the packages.
Create the build set tar.
A package tar file can be created by adding the --pkg-tar-files to the sb-set-builder command. This creates a tar file per package built in the build set:
$ ../source-builder/sb-set-builder --log=l-sparc.txt \
--prefix=$HOME/development/rtems/4.11 \
--bset-tar-file \
--pkg-tar-files \ <1>
--no-install 4.11/rtems-sparc
Source Builder - Set Builder, v0.2.0
Build Set: 4.11/rtems-sparc
config: expat-2.1.0-1.cfg
package: expat-2.1.0-x86_64-freebsd9.1-1
building: expat-2.1.0-x86_64-freebsd9.1-1
config: tools/rtems-binutils-2.22-1.cfg
package: sparc-rtems4.11-binutils-2.22-1
building: sparc-rtems4.11-binutils-2.22-1
config: tools/rtems-gcc-4.7.2-newlib-1.20.0-1.cfg
package: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
building: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
config: tools/rtems-gdb-7.5.1-1.cfg
package: sparc-rtems4.11-gdb-7.5.1-1
building: sparc-rtems4.11-gdb-7.5.1-1
tarball: tar/rtems-4.11-sparc-rtems4.11-1.tar.bz2
cleaning: expat-2.1.0-x86_64-freebsd9.1-1
cleaning: sparc-rtems4.11-binutils-2.22-1
cleaning: sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1
cleaning: sparc-rtems4.11-gdb-7.5.1-1
Build Set: Time 0:14:37.658460
$ ls tar
expat-2.1.0-x86_64-freebsd9.1-1.tar.bz2 sparc-rtems4.11-binutils-2.22-1.tar.bz2
sparc-rtems4.11-gdb-7.5.1-1.tar.bz2 <2> rtems-4.11-sparc-rtems4.11-1.tar.bz2 <3>
sparc-rtems4.11-gcc-4.7.2-newlib-1.20.0-1.tar.bz2
Items
The option to create packages tar files.
The GDB package tar file.
The build set tar file. All the others in a single tar file.
13.8. Bugs, Crashes, and Build Failures¶
The RTEMS Source Builder is a Python program and every care is taken to test the code however bugs, crashes, and build failures can and do happen. If you find a bug please report it via the Developer Site or email on the RTEMS Users list.
Please include the generated RSB report. If you see the following a report has been generated:
...
...
Build FAILED <1>
See error report: rsb-report-4.11-rtems-lm32.txt <2>
Items:
The build has failed.
The report’s file name.
The generated report contains the command line, version of the RSB, your host’s uname details, the version of Python and the last 200 lines of the log.
If for some reason there is no report please send please report the following:
Command line,
The git hash,
Host details with the output of the
uname -acommand,If you have made any modifications.
If there is a Python crash please cut and paste the Python backtrace into the bug report. If the tools fail to build please locate the first error in the log file. This can be difficult to find on hosts with many cores so it sometimes pays to re-run the command with the --jobs=none option to get a log that is correctly sequenced. If searching the log file seach for error: and the error should be just above it.
13.8.1. Contributing¶
We welcome all users adding, fixing, updating and upgrading packages and their configurations. The RSB is open source and open to contributions. These can be bug fixes, new features or new configurations. Please break patches down into changes to the core Python code, configuration changes or new configurations.
Please email patches generated using git so your commit messages and you are acknowledged as the contributor.
13.9. History¶
The RTEMS Source Builder is a stand alone tool based on another tool called the SpecBuilder written by Chris Johns. The SpecBuilder was written around 2010 for the RTEMS project to provide Chris with a way to build tools on hosts that did not support RPMs. At the time the RTEMS tools maintainer only supported spec files and these files held all the vital configuration data needed to create suitable tool sets. The available SRPM and spec files by themselves where of little use because a suitable rpm tool was needed to use them. At the time the available versions of rpm for a number of non-RPM hosts were broken and randomly maintained. The solution Chris settled on was to use the spec files and to write a Python based tool that parsed the spec file format creating a shell script that could be run to build the package. The approach proved successful and Chris was able to track the RPM version of the RTEMS tools on a non-RPM host for a number of years.
The SpecBuilder tool did not build tools or packages unrelated to the RTEMS Project where no suitable spec file was available so another tool was needed. Rather than start again Chris decided to take the parsing code for the spec file format and build a new tool called the RTEMS Source Builder.
14. Glossary¶
- ABI
Application Binary Interface
- Architecture
Family or class of processor based around a common instruction set. RTEMS architectures follow the GCC architecture model as RTEMS needs an GCC architecture compiler for each support RTEMS architecture.
- APA
Arbitrary Processor Affinity
- API
Application Programming Interface
- Binutils
GNU Binary Utilities such as the assembler
as, linkerldand a range of other tools used in the development of software.- BSP
Board Support Package is a specific configuration RTEMS can be built for. An RTEMS install process installs specific library and headers files for a single BSP. A BSP optimises RTEMS to a specific target hardware.
- Buildbot
A continuous inteagration build server.
- C11
ISO/IEC 9899:2011
- C++11
ISO/IEC 14882:2011
- Crosscompiler
A compiler built to run on a Host that generate code for another architecture.
- DLL
Dynamically Linker Library used on Windows.
- EDF
Earliest Deadline First
- EMB²
- FAT
File Allocation Table
- Futex
Fast User-Space Locking
- IMFS
In-Memory File System
- JFFS2
Journalling Flash File System version 2
- GCC
GNU Compiler Collection
- GDB
GNU Debugger
- GNU
GNU’s Not Unix
- Host
The computer and operating system that hosts the RTEMS development tools such as the compiler, linker and debugger.
- MinGW
Minimal GNU system for Windows that lets GCC built programs use the standard Windows operating system DLLs. It lets you build native Windows programs with the GNU GCC compiler.
- MinGW64
Minimal GNU system for 64bit Windows. MinGW64 is not the MinGW project.
- MrsP
Multiprocessor Resource-Sharing Protocol
- MSYS2
Minimal System 2 is a fork of the MinGW project’s MSYS tool and the MinGW MSYS tool is a fork of Cygwin project. The Cygwin project provides a POSIX emulation layer for Windows so POSIX software can run on Windows. MSYS is a minimal version that is just enough to let
configurescripts run. MSYS has a simplified path structure to make it easier to building native Windows programs.- NFSv2
Network File System version 2
- OMIP
\(O(m)\) Independence-Preserving Protocol
- OpenMP
Open Multi-Processing
- POSIX
Portable Operating System Interface is a standard that lets software be portable between compliant operating systems.
- prefix
A path used when building a package so all parts of the package reside under that path.
- RFS
RTEMS File System
- RSB
RTEMS Source Builder is part of the RTEMS Tools Project. It builds packages such as the tools for the RTEMS operating system.
- RTEMS
The Real-Time Executive for Multiprocessor Systems or RTEMS is an open source fully featured Real Time Operating System or RTOS that supports a variety of open standard application programming interfaces (API) and interface standards such as POSIX and BSD sockets.
- SMP
Symmetric Multiprocessing
- Target
A target is the hardware or simulator a BSP built executable runs on.
- Test Suite
See Testsuite
- Testsuite
RTEMS test suite located in the
testsuites/directory.- TLS
Thread-Local Storage
- Waf
Waf build system. For more information see http://www.waf.io/
- YAFFS2