Cross Compiling Options

What is your strategy on cross compilation toolchains? Here we explore a
few options:

  • DIY - requires you to manually start each step, unless you're comfortable enough to script the process. As each step takes ages, this can be a tedious job. However, it gives you full control and insight into each step.
  • crosstool [http://www.kegel.com/crosstool/] - a scripted version of DIY + patches. Used to be my preferred tool, but hasn't been updated since 2006. Limited to glibc.
  • crosstool-NG [http://ymorin.is-a-geek.org/projects/crosstool] - a rewrite of crosstool. Provides a menuconfig configuration interface, which makes it quite easy to use. You can also pass configurations around. Supports glibc, but also uclibc and eglibc.
  • buildroot [http://buildroot.uclibc.org/] - builds a uclibc-based gcc in the process of building an entire system. The toolchain can be ripped out (you can also force buildroot to use an external toolchain - thus avoiding building the same toolchain over and over again). As the focus is on the entire system, this approach provides little insight of the toolchain, but in my experience, the outcome usually works well.
  • scratchbox [http://www.scratchbox.org/] - provides uclibc or glibc based toolchains. Approaches cross compilation is a new way - providing an environment where the actual cross part of the compilation is abstracted away. Also provides an emulated target environment, so you can run your ARM binaries on your standard PC from within the scratchbox environment.

The available options are ordered in some sort of control freak factor order, where DIY provides loads of control over every tiny aspect, while scratchbox just makes things work.

Myself? Last time around I used ct-ng, but I've used them all (and will not DIY again unless forced ;) ).

______________________

Johan Thelin is a consultant working with Qt, embedded and free
software. On-line, he is known as e8johan.

Comments

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

In my opinion, Openembedded

Anonymous's picture

In my opinion, Openembedded is also a good option to be mentioned here? With openembedded you can generate installable cross compiler toolchains for many target architectures, besides building complete target images etc. It is very configurable.

I still use crosstool

Vidyasagara Guntaka's picture

Kegel's crosstool is still my favorite. Last week I modified it to build absolute latest toolchain combo GCC 4.5.0 + GLIBC 2.11.2 + BINUTILS 2.20 for powerpc32.

Crossplex

Laurent Parenteau's picture

I've recently been introduced to Crossplex. While I haven't tried it yet, it looks promising.

MinGW

Johan Thelin's picture

@BaloneyGeek - you are completely right. As an embedded guy, I kind of completely forgot the OS-to-OS cross compilation needs - especially MinGW. Briefly googling the topic, I ran into a thread where the CT-NG people are discussing adding mingw as a kernel type, thus allowing the build of Windows compatible toolchains from CT-NG.

Johan Thelin is a consultant working with Qt, embedded and free
software. On-line, he is known as e8johan.

Personally I used crosstool

ciotog's picture

Personally I used crosstool to build a toolchain to build applications that will run on Lantronix's SecureLinx console server (it uses a MIPS-EL processor with glibc - Debian Sarge packages are compatible). For the most part it was a matter of selecting the libc library and cpu type, then sit back and let it build.

For a ppc - uclibc embedded product we have at work I simply used ELDK 4.1 by DENX, which has a complete pre-compiled cross toolchain as well as a root filesystem with a complete build environment for native compilation for the target (if needed). Couldn't get any simpler than that!

DIY is the only way for certain things.

@BaloneyGeek's picture

Suppose you were building the MinGW32 or MinGW64 toolchain. DIY would be the only way, unless you wanted to installl your distro packages, which'd always be somehow half broken anyway, or use the mingwmsvc- target which never seems to work out.

Webcast
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers

Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.

Learn More

Sponsored by AMD

White Paper
Red Hat White Paper: Using an Open Source Framework to Catch the Bad Guy

Built-in forensics, incident response, and security with Red Hat Enterprise Linux 6

Every security policy provides guidance and requirements for ensuring adequate protection of information and data, as well as high-level technical and administrative security requirements for a system in a given environment. Traditionally, providing security for a system focuses on the confidentiality of the information on it. However, protecting the data integrity and system and data availability is just as important. For example, when processing United States intelligence information, there are three attributes that require protection: confidentiality, integrity, and availability.

Learn more about catching the bad guy in this free white paper.

Learn More

Sponsored by DLT Solutions