Back: Package Installation
Forward: Host and Target
FastBack: Invoking configure
Up: Top
FastForward: Installing GNU Autotools
Top: Autoconf, Automake, and Libtool
Contents: Table of Contents
Index: Index
About: About this document

26. Cross Compilation with GNU Autotools

Normally, when you build a program, it runs on the system on which you built it. For example, if you compile a simple program, you can immediately run it on the same machine.

This is normally how GNU Autotools is used as well. You run the `configure' script on a particular machine, you run make on the same machine, and the resulting program also runs on the same machine. However, there are cases where it is useful to build a program on one machine and run it on another.

One common example is a program which runs on an embedded system. An embedded system is a special purpose computer, often part of a larger system, such as the computers found within modern automobiles. An embedded system often does not support a general programming environment, so there is no way to run a shell or a compiler on the embedded system. However, it is still necessary to write programs to run on the embedded system. These programs are built on a different machine, normally a general purpose computer. The resulting programs can not be run directly on the general purpose computer. Instead, they are copied onto the embedded system and run there. (We are omitting many details and possibilities of programming embedded systems here, but this should be enough to understand the the points relevant to GNU Autotools. For more information, see a book such as Programming Embedded Systems by Michael Barr.)

Another example where it is useful to build a program on one machine and run it on another is the case when one machine is much faster. It can sometimes be useful to use the faster machine as a compilation server, to build programs which are then copied to the slower machine and run there.

Building a program on one type of system which runs on a different type of system is called cross compiling. Doing this requires a specially configured compiler, known as a cross compiler. Similarly, we speak of cross assemblers, cross linkers, etc. When it is necessary to explicitly distinguish the ordinary sort of compiler, whose output runs on the same type of system, from a cross compiler, we call the ordinary compiler a native compiler. Although the debugger is not strictly speaking a compilation tool, it is meaningful to speak of a cross debugger: a debugger which is used to debug code which runs on another system.

GNU Autotools supports cross compilation in two distinct though related ways. Firstly, GNU Autotools supports configuring and building a cross compiler or other cross compilation tools. Secondly, GNU Autotools supports building tools using a cross compiler (this is sometimes called a Canadian Cross). In the rest of this chapter we will explain how to use GNU Autotools to do these tasks.

If you are not interested in doing cross compilation, you may skip this chapter. However, if you are developing `configure' scripts, we recommend that you at least skim this chapter to get some hints as to how to write them so that it is possible to build your package using a cross compiler; in particular, see 26.4.6 Supporting Building with a Cross Compiler. Even if your package is useless for an embedded system, it is possible that somebody with a very fast compilation server will want to use it to cross compile your package.

This document was generated by Gary V. Vaughan on February, 8 2006 using texi2html