1 File.........: overview.txt
2 Content......: Overview of how crosstool-NG works.
3 Copyrigth....: (C) 2007 Yann E. MORIN <yann.morin.1998@anciens.enib.fr>
4 License......: Creative Commons Attribution Share Alike (CC-by-sa), v2.5
14 Installing crosstool-NG
17 Preparing for packaging
20 Configuring crosstool-NG
21 Interesting config options
22 Re-building an existing toolchain
24 Stopping and restarting a build
25 Testing all toolchains at once
26 Overriding the number of // jobs
32 Seemingly-native toolchains
40 Adding a new version of a component
49 crosstool-NG aims at building toolchains. Toolchains are an essential component
50 in a software development project. It will compile, assemble and link the code
51 that is being developed. Some pieces of the toolchain will eventually end up
52 in the resulting binary/ies: static libraries are but an example.
54 So, a toolchain is a very sensitive piece of software, as any bug in one of the
55 components, or a poorly configured component, can lead to execution problems,
56 ranging from poor performance, to applications ending unexpectedly, to
57 mis-behaving software (which more than often is hard to detect), to hardware
58 damage, or even to human risks (which is more than regrettable).
60 Toolchains are made of different piece of software, each being quite complex
61 and requiring specially crafted options to build and work seamlessly. This
62 is usually not that easy, even in the not-so-trivial case of native toolchains.
63 The work reaches a higher degree of complexity when it comes to cross-
64 compilation, where it can become quite a nightmare...
66 Some cross-toolchains exist on the internet, and can be used for general
67 development, but they have a number of limitations:
68 - they can be general purpose, in that they are configured for the majority:
69 no optimisation for your specific target,
70 - they can be prepared for a specific target and thus are not easy to use,
71 nor optimised for, or even supporting your target,
72 - they often are using aging components (compiler, C library, etc...) not
73 supporting special features of your shiny new processor;
74 On the other side, these toolchain offer some advantages:
75 - they are ready to use and quite easy to install and setup,
76 - they are proven if used by a wide community.
78 But once you want to get all the juice out of your specific hardware, you will
79 want to build your own toolchain. This is where crosstool-NG comes into play.
81 There are also a number of tools that build toolchains for specific needs,
82 which are not really scalable. Examples are:
83 - buildroot (buildroot.uclibc.org) whose main purpose is to build root file
84 systems, hence the name. But once you have your toolchain with buildroot,
85 part of it is installed in the root-to-be, so if you want to build a whole
86 new root, you either have to save the existing one as a template and
87 restore it later, or restart again from scratch. This is not convenient,
88 - ptxdist (www.pengutronix.de/software/ptxdist), whose purpose is very
90 - other projects (openembedded.org for example), which are again used to
91 build root file systems.
93 crosstool-NG is really targeted at building toolchains, and only toolchains.
94 It is then up to you to use it the way you want.
102 crosstool was first 'conceived' by Dan Kegel, who offered it to the community
103 as a set of scripts, a repository of patches, and some pre-configured, general
104 purpose setup files to be used to configure crosstool. This is available at
105 http://www.kegel.com/crosstool, and the subversion repository is hosted on
106 google at http://code.google.com/p/crosstool/.
108 I once managed to add support for uClibc-based toolchains, but it did not make
109 into mainline, mostly because I didn't have time to port the patch forward to
110 the new versions, due in part to the big effort it was taking.
112 So I decided to clean up crosstool in the state it was, re-order the things
113 in place, add appropriate support for what I needed, that is uClibc support
114 and a menu-driven configuration, named the new implementation crosstool-NG,
115 (standing for crosstool Next Generation, as many other comunity projects do,
116 and as a wink at the TV series "Star Trek: The Next Generation" ;-) ) and
117 made it available to the community, in case it was of interest to any one.
120 ___________________________
122 Installing crosstool-NG /
123 ________________________/
125 There are two ways you can use crosstool-NG:
126 - build and install it, then get rid of the sources like you'd do for most
128 - or only build it and run from the source directory.
130 The former should be used if you got crosstool-NG from a packaged tarball, see
131 "Install method", below, while the latter is most useful for developpers that
132 use a clone of the repository, and want to submit patches, see "The Hacker's
138 If you go for the install, then you just follow the classical, but yet easy
140 ./configure --prefix=/some/place
143 export PATH="${PATH}:/some/place/bin"
145 You can then get rid of crosstool-NG source. Next create a directory to serve
146 as a working place, cd in there and run:
149 See below for complete usage.
154 If you go the hacker's way, then the usage is a bit different, although very
159 Now, *do not* remove crosstool-NG sources. They are needed to run crosstool-NG!
160 Stay in the directory holding the sources, and run:
163 See below for complete usage.
165 Now, provided you used a clone of the repository, you can send me your changes.
166 See the section titled CONTRIBUTING, below, for how to submit changees.
168 Preparing for packaging |
169 ------------------------+
171 If you plan on packaging crosstool-NG, you surely don't want to install it
172 in your root file system. The install procedure of crosstool-NG honors the
175 ./configure --prefix=/usr
177 make DESTDIR=/packaging/place install
182 crosstool-NG comes with a shell script fragment that defines bash-compatible
183 completion. That shell fragment is currently not installed automatically, but
186 To install the shell script fragment, you have two options:
187 - install system-wide, most probably by copying ct-ng.comp into
188 /etc/bash_completion.d/
189 - install for a single user, by copying ct-ng.comp into ${HOME}/ and
190 sourcing this file from your ${HOME}/.bashrc
195 Some people contibuted code that couldn't get merged for various reasons. This
196 code is available as lzma-compressed patches, in the contrib/ sub-directory.
197 These patches are to be applied to the source of crosstool-NG, prior to
198 installing, using something like the following:
199 lzcat contrib/foobar.patch.lzma |patch -p1
201 There is no guarantee that a particuliar contribution applies to the current
202 version of crosstool-ng, or that it will work at all. Use contributions at
206 ____________________________
208 Configuring crosstool-NG /
209 _________________________/
211 crosstool-NG is configured with a configurator presenting a menu-stuctured set
212 of options. These options let you specify the way you want your toolchain
213 built, where you want it installed, what architecture and specific processor it
214 will support, the version of the components you want to use, etc... The
215 value for those options are then stored in a configuration file.
217 The configurator works the same way you configure your Linux kernel. It is
218 assumed you now how to handle this.
220 To enter the menu, type:
223 Almost every config item has a help entry. Read them carefully.
225 String and number options can refer to environment variables. In such a case,
226 you must use the shell syntax: ${VAR}. You shall neither single- nor double-
227 quote the string/number options.
229 There are three environment variables that are computed by crosstool-NG, and
233 It represents the target tuple you are building for. You can use it for
234 example in the installation/prefix directory, such as:
235 /opt/x-tools/${CT_TARGET}
238 The top directory where crosstool-NG is running. You shouldn't need it in
239 most cases. There is one case where you may need it: if you have local
240 patches and you store them in your running directory, you can refer to them
241 by using CT_TOP_DIR, such as:
242 ${CT_TOP_DIR}/patches.myproject
245 The version of crosstool-NG you are using. Not much use for you, but it's
246 there if you need it.
248 Interesting config options |
249 ---------------------------+
251 CT_LOCAL_TARBALLS_DIR:
252 If you already have some tarballs in a direcotry, enter it here. That will
253 speed up the retrieving phase, where crosstool-NG would otherwise download
257 This is where the toolchain will be installed in (and for now, where it
258 will run from). Common use is to add the target tuple in the directory
259 path, such as (see above):
260 /opt/x-tools/${CT_TARGET}
263 An identifier for your toolchain, will take place in the vendor part of the
264 target tuple. It shall *not* contain spaces or dashes. Usually, keep it
265 to a one-word string, or use underscores to separate words if you need.
266 Avoid dots, commas, and special characters.
269 An alias for the toolchian. It will be used as a prefix to the toolchain
270 tools. For example, you will have ${CT_TARGET_ALIAS}-gcc
272 Also, if you think you don't see enough versions, you can try to enable one of
276 Show obsolete versions or tools. Most of the time, you don't want to base
277 your toolchain on too old a version (of gcc, for example). But at times, it
278 can come handy to use such an old version for regression tests. Those old
279 versions are hidden behind CT_OBSOLETE. Those versions (or features) are so
280 marked because maintaining support for those in crosstool-NG would be too
281 costly, time-wise, and time is dear.
284 Show experimental versions or tools. Again, you might not want to base your
285 toolchain on too recent tools (eg. gcc) for production. But if you need a
286 feature present only in a recent version, or a new tool, you can find them
287 hidden behind CT_EXPERIMENTAL. Those versions (or features) did not (yet)
288 receive thorough testing in crosstool-NG, and/or are not mature enough to
291 Re-building an existing toolchain |
292 ----------------------------------+
294 If you have an existing toolchain, you can re-use the options used to build it
295 to create a new toolchain. That needs a very little bit of effort on your side
296 but is quite easy. The options to build a toolchain are saved with the
297 toolchain, and you can retrieve this configuration by running:
300 This will dump the configuration to stdout, so to rebuild a toolchain with this
301 configuration, the following is all you need to do:
302 ${CT_TARGET}-config >.config
305 Then, you can review and change the configuration by running:
309 ________________________
311 Running crosstool-NG /
312 _____________________/
314 To build the toolchain, simply type:
317 This will use the above configuration to retrieve, extract and patch the
318 components, build, install and eventually test your newly built toolchain.
320 You are then free to add the toolchain /bin directory in your PATH to use
323 In any case, you can get some terse help. Just type:
328 Stopping and restarting a build |
329 --------------------------------+
331 If you want to stop the build after a step you are debugging, you can pass the
332 variable STOP to make:
335 Conversely, if you want to restart a build at a specific step you are
336 debugging, you can pass the RESTART variable to make:
337 ct-ng RESTART=some_step
339 Alternatively, you can call make with the name of a step to just do that step:
342 ct-ng RESTART=libc_headers STOP=libc_headers
344 The shortcuts +step_name and step_name+ allow to respectively stop or restart
346 ct-ng +libc_headers and: ct-ng libc_headers+
348 ct-ng STOP=libc_headers and: ct-ng RESTART=libc_headers
350 To obtain the list of acceptable steps, please call:
353 Note that in order to restart a build, you'll have to say 'Y' to the config
354 option CT_DEBUG_CT_SAVE_STEPS, and that the previous build effectively went
357 Building all toolchains at once |
358 --------------------------------+
360 You can build all samples; simply call:
363 Overriding the number of // jobs |
364 ---------------------------------+
366 If you want to override the number of jobs to run in // (the -j option to
367 make), you can either re-enter the menuconfig, or simply add it on the command
371 which tells crosstool-NG to override the number of // jobs to 4.
373 You can see the actions that support overriding the number of // jobs in
374 the help menu. Those are the ones with [.#] after them (eg. build[.#] or
375 build-all[.#], and so on...).
380 The crosstool-NG script 'ct-ng' is a Makefile-script. It does *not* execute
381 in parallel (there is not much to gain). When speaking of // jobs, we are
382 refering to the number of // jobs when making the *components*. That is, we
383 speak of the number of // jobs used to build gcc, glibc, and so on...
388 Starting with gcc-4.3 come two new dependencies: GMP and MPFR. With gcc-4.4,
389 come three new ones: GMP, PPL and CLooG/ppl. These are libraries that enable
390 advanced features to gcc. Additionally, some of the libraries can be used by
391 binutils and gdb. Unfortunately, not all systems on which crosstool-NG runs
392 have all of those libraries. And for those that do, the versions of those
393 libraries may be older than the version required by gcc.
395 This is why crosstool-NG builds its own set of libraries as part of the
398 The libraries are built as shared libraries, because building them as static
399 libraries has some short-comings. This poses no problem at build time, as
400 crosstool-NG correctly points gcc (and binutils and gdb) to the correct
401 place where our own version of the libraries are installed. But it poses
402 a problem when gcc et al. are run: the place where the libraries are is most
403 probably not known to the host dynamic linker. Still worse, if the host system
404 has its own versions, then ld.so would load the wrong library!
406 So we have to force the dynamic linker to load the correct version. We do this
407 by using the LD_LIBRARY_PATH variable, that informs the dynamic linker where
408 to look for shared libraries prior to searching its standard places. But we
409 can't impose that burden on all the system (because it'd be a nightmare to
410 configure, and because two toolchains on the same system may use different
411 versions of the libraries); so we have to do it on a per-toolchain basis.
413 So we rename all binaries of the toolchain (by adding a dot '.' as their first
414 character), and add a small program, the so-called "tools wrapper", that
415 correctly sets LD_LIBRARY_PATH prior to running the real tool.
417 First, the wrapper was written as a POSIX-compliant shell script. That shell
418 script is very simple, if not trivial, and works great. The only drawback is
419 that it does not work on host systems that lack a shell, for example the
420 MingW32 environment. To solve the issue, the wrapper has been re-written in C,
421 and compiled at build time. This C wrapper is much more complex than the shell
422 script, and although it sems to be working, it's been only lightly tested.
423 Some of the expected short-comings with this C wrapper are;
424 - multi-byte file names may not be handled correctly
425 - it's really big for what it does
427 So, the default wrapper installed with your toolchain is the shell script.
428 If you know that your system is missing a shell, then you shall use the C
429 wrapper (and report back whether it works, or does not work, for you).
432 _______________________
434 Using the toolchain /
435 ____________________/
437 Using the toolchain is as simple as adding the toolchain's bin directory in
439 export PATH="${PATH}:/your/toolchain/path/bin"
441 and then using the target tuple to tell the build systems to use your
443 ./configure --target=your-target-tuple
445 make CC=your-target-tuple-gcc
447 make CROSS_COMPILE=your-target-tuple-
450 It is strongly advised not to use the toolchain sys-root directory as an
451 install directory for your programs/packages. If you do so, you will not be
452 able to use your toolchain for another project. It is even strongly advised
453 that your toolchain is chmod-ed to read-only once successfully build, so that
454 you don't go polluting your toolchain with your programs/packages' files.
456 Thus, when you build a program/package, install it in a separate directory,
457 eg. /your/root. This directory is the /image/ of what would be in the root file
458 system of your target, and will contain all that your programs/packages have
461 The 'populate' script |
462 ----------------------+
464 When your root directory is ready, it is still missing some important bits: the
465 toolchain's libraries. To populate your root directory with those libs, just
467 your-target-tuple-populate -s /your/root -d /your/root-populated
469 This will copy /your/root into /your/root-populated, and put the needed and only
470 the needed libraries there. Thus you don't polute /your/root with any cruft that
471 would no longer be needed should you have to remove stuff. /your/root always
472 contains only those things you install in it.
474 You can then use /your/root-populated to build up your file system image, a
475 tarball, or to NFS-mount it from your target, or whatever you need.
477 The populate script accepts the following options:
480 Use 'src_dir' as the un-populated root directory.
483 Put the populated root directory in 'dst_dir'.
486 Always add specified libraries.
489 Always add libraries listed in 'file'.
492 Remove 'dst_dir' if it previously existed; continue even if any library
493 specified with -l or -L is missing.
496 Be verbose, and tell what's going on (you can see exactly where libs are
502 See 'your-target-tuple-populate -h' for more information on the options.
504 Here is how populate works:
506 1) performs some sanity checks:
507 - src_dir and dst_dir are specified
509 - unless forced, dst_dir does not exist
512 2) copy src_dir to dst_dir
514 3) add forced libraries to dst_dir
515 - build the list from -l and -L options
516 - get forced libraries from the sysroot (see below for heuristics)
517 - abort on the first missing library, unless -f is specified
519 4) add all missing libraries to dst_dir
520 - scan dst_dir for every ELF files that are 'executable' or
522 - list the "NEEDED Shared library" fields
523 - check if the library is already in dst_dir/lib or dst_dir/usr/lib
524 - if not, get the library from the sysroot
525 - if it's in sysroot/lib, copy it to dst_dir/lib
526 - if it's in sysroot/usr/lib, copy it to dst_dir/usr/lib
527 - in both cases, use the SONAME of the library to create the file
529 - if it was not found in the sysroot, this is an error.
537 There are four kinds of toolchains you could encounter.
539 First off, you must understand the following: when it comes to compilers there
540 are up to four machines involved:
541 1) the machine configuring the toolchain components: the config machine
542 2) the machine building the toolchain components: the build machine
543 3) the machine running the toolchain: the host machine
544 4) the machine the toolchain is generating code for: the target machine
546 We can most of the time assume that the config machine and the build machine
547 are the same. Most of the time, this will be true. The only time it isn't
548 is if you're using distributed compilation (such as distcc). Let's forget
549 this for the sake of simplicity.
551 So we're left with three machines:
556 Any toolchain will involve those three machines. You can be as pretty sure of
557 this as "2 and 2 are 4". Here is how they come into play:
559 1) build == host == target
560 This is a plain native toolchain, targetting the exact same machine as the
561 one it is built on, and running again on this exact same machine. You have
562 to build such a toolchain when you want to use an updated component, such
563 as a newer gcc for example.
564 crosstool-NG calls it "native".
566 2) build == host != target
567 This is a classic cross-toolchain, which is expected to be run on the same
568 machine it is compiled on, and generate code to run on a second machine,
570 crosstool-NG calls it "cross".
572 3) build != host == target
573 Such a toolchain is also a native toolchain, as it targets the same machine
574 as it runs on. But it is build on another machine. You want such a
575 toolchain when porting to a new architecture, or if the build machine is
576 much faster than the host machine.
577 crosstool-NG calls it "cross-native".
579 4) build != host != target
580 This one is called a canadian-toolchain (*), and is tricky. The three
581 machines in play are different. You might want such a toolchain if you
582 have a fast build machine, but the users will use it on another machine,
583 and will produce code to run on a third machine.
584 crosstool-NG calls it "canadian".
586 crosstool-NG can build all these kinds of toolchains (or is aiming at it,
589 (*) The term Canadian Cross came about because at the time that these issues
590 were all being hashed out, Canada had three national political parties.
591 http://en.wikipedia.org/wiki/Cross_compiler
593 Seemingly-native toolchains |
594 ----------------------------+
596 Seemingly-native toolchains are toolchains that target the same architecture
597 as the one it is built on, and on which it will run, but the machine tuple
598 may be different (eg i686 vs. i386, or x86_64-unknown-linux-gnu vs.
599 x86_64-pc-linux-gnu). This also applies if the target architecture is of the
600 same kind (eg. x86 vs. x86_64, or ppc vs. ppc64).
602 Such toolchain is tricky to build, as the configure scripts may incorrectly
603 assume that files (headers and libs) from the build (or host) machine can be
604 used by the cross-compiler it is going to build. The problem seems to arise
605 only with glibc (and eglibc?) starting with version 2.7.
613 Sending a bug report |
614 ---------------------+
616 If you need to send a bug report, please send a mail with subject
617 prefixed with "[CT_NG]" with to following destinations:
618 TO: yann.morin.1998 (at) anciens.enib.fr
619 CC: crossgcc (at) sourceware.org
624 If you want to enhance crosstool-NG, there's a to-do list in the TODO file.
626 Patches should come with the appropriate SoB line. A SoB line is typically
628 Signed-off-by: John DOE <john.doe@somewhere.net>
630 The SoB line is clearly described in Documentation/SubmittingPatches , section
631 12, of your favourite Linux kernel source tree.
633 Then you'll need to correctly configure Mercurial. There are two extensions
634 that you may find usefull:
635 - mq : http://mercurial.selenic.com/wiki/MqExtension
636 - patchbomb : http://mercurial.selenic.com/wiki/PatchbombExtension
638 Commit messages should look like (without leading pipes):
639 |component: short, one-line description
641 |optional longer description
642 |on multiple lines if needed
644 Here is an example commit message (see revision a53a5e1d61db):
645 |comp-libs/cloog: fix building
647 |For CLooG/PPL 0.15.3, the directory name was simply cloog-ppl.
648 |For any later versions, the directory name does have the version, such as
651 Here's a typical hacking session:
652 hg clone http://ymorin.is-a-geek.org/hg/crosstool-ng crosstool-ng
655 hg qnew -D -U -e my_first_patch
656 *edit patch description*
657 *hack* *hack* *check* *fails* *hack* *hack* *check* *works*
659 *edit patch description, serving as commit message*
660 hg qnew -D -U -e my_second_patch
661 *edit patch description*
662 *hack* *hack* *check* *fails* *hack* *hack* *check* *works*
664 *edit patch description, serving as commit message*
665 hg email --outgoing --intro \
666 --from '"Your Full NAME" <your.email (at) your.domain>' \
667 --to '"Yann E. MORIN" <yann.morin.1998 (at) anciens.enib.fr>' \
668 --cc 'crossgcc (at) sourceware.org'
669 *edit introductory message*
671 *re-send if no answer for a few days*
673 Note: replace '(at)' above with a plain '@'.
681 Internally, crosstool-NG is script-based. To ease usage, the frontend is
687 The entry point to crosstool-NG is the Makefile script "ct-ng". Calling this
688 script with an action will act exactly as if the Makefile was in the current
689 working directory and make was called with the action as rule. Thus:
692 is equivalent to having the Makefile in CWD, and calling:
695 Having ct-ng as it is avoids copying the Makefile everywhere, and acts as a
698 ct-ng loads sub- Makefiles from the library directory $(CT_LIB_DIR), as set up
699 at configuration time with ./configure.
701 ct-ng also searches for config files, sub-tools, samples, scripts and patches in
702 that library directory.
704 Because of a stupid make behavior/bug I was unable to track down, implicit make
705 rules are disabled: installing with --local would triger those rules, and mconf
711 The kconfig language is a hacked version, vampirised from the Linux kernel
712 (http://www.kernel.org/), and (heavily) adapted to my needs.
714 The list of the most notable changes (at least the ones I remember) follows:
715 - the CONFIG_ prefix has been replaced with CT_
716 - a leading | in prompts is skipped, and subsequent leading spaces are not
718 - otherwise leading spaces are silently trimmed
720 The kconfig parsers (conf and mconf) are not installed pre-built, but as
721 source files. Thus you can have the directory where crosstool-NG is installed,
722 exported (via NFS or whatever) and have clients with different architectures
723 use the same crosstool-NG installation, and most notably, the same set of
726 Architecture-specific |
727 ----------------------+
729 Note: this chapter is not really well written, and might thus be a little bit
730 complex to understand. To get a better grasp of what an architecture is, the
731 reader is kindly encouraged to look at the "arch/" sub-directory, and to the
732 existing architectures to see how things are laid out.
734 An architecture is defined by:
736 - a human-readable name, in lower case letters, with numbers as appropriate.
737 The underscore is allowed; space and special characters are not.
739 - a file in "config/arch/", named after the architecture's name, and suffixed
741 Eg.: config/arch/arm.in
742 - a file in "scripts/build/arch/", named after the architecture's name, and
744 Eg.: scripts/build/arch/arm.sh
746 The architecture's ".in" file API:
747 > the config option "ARCH_%arch%" (where %arch% is to be replaced with the
748 actual architecture name).
749 That config option must have *neither* a type, *nor* a prompt! Also, it can
750 *not* depend on any other config option (EXPERIMENTAL is managed as above).
754 defines a (terse) help entry for this architecture:
758 The ARM architecture.
760 selects adequate associated config options.
761 Note: 64-bit architectures *shall* select ARCH_64
764 select ARCH_SUPPORTS_BOTH_ENDIAN
765 select ARCH_DEFAULT_LE
767 The ARM architecture.
772 The x86_64 architecture.
774 > other target-specific options, at your discretion. Note however that to
775 avoid name-clashing, such options shall be prefixed with "ARCH_%arch%",
776 where %arch% is again replaced by the actual architecture name.
777 (Note: due to historical reasons, and lack of time to clean up the code,
778 I may have left some config options that do not completely conform to
779 this, as the architecture name was written all upper case. However, the
780 prefix is unique among architectures, and does not cause harm).
782 The architecture's ".sh" file API:
783 > the function "CT_DoArchTupleValues"
786 - all variables from the ".config" file,
787 - the two variables "target_endian_eb" and "target_endian_el" which are
788 the endianness suffixes
789 + return value: 0 upon success, !0 upon failure
792 - the environment variable CT_TARGET_ARCH
794 the architecture part of the target tuple.
795 Eg.: "armeb" for big endian ARM
799 - the environment variable CT_TARGET_SYS
801 the sytem part of the target tuple.
802 Eg.: "gnu" for glibc on most architectures
803 "gnueabi" for glibc on an ARM EABI
805 - for glibc-based toolchain: "gnu"
806 - for uClibc-based toolchain: "uclibc"
809 - the environment variable CT_KERNEL_ARCH
811 the architecture name as understandable by the Linux kernel build
813 Eg.: "arm" for an ARM
814 "powerpc" for a PowerPC
820 - the environment variables to configure the cross-gcc (defaults)
821 - CT_ARCH_WITH_ARCH : the gcc ./configure switch to select architecture level ( "--with-arch=${CT_ARCH_ARCH}" )
822 - CT_ARCH_WITH_ABI : the gcc ./configure switch to select ABI level ( "--with-abi=${CT_ARCH_ABI}" )
823 - CT_ARCH_WITH_CPU : the gcc ./configure switch to select CPU instruction set ( "--with-cpu=${CT_ARCH_CPU}" )
824 - CT_ARCH_WITH_TUNE : the gcc ./configure switch to select scheduling ( "--with-tune=${CT_ARCH_TUNE}" )
825 - CT_ARCH_WITH_FPU : the gcc ./configure switch to select FPU type ( "--with-fpu=${CT_ARCH_FPU}" )
826 - CT_ARCH_WITH_FLOAT : the gcc ./configure switch to select floating point arithmetics ( "--with-float=soft" or /empty/ )
829 - the environment variables to pass to the cross-gcc to build target binaries (defaults)
830 - CT_ARCH_ARCH_CFLAG : the gcc switch to select architecture level ( "-march=${CT_ARCH_ARCH}" )
831 - CT_ARCH_ABI_CFLAG : the gcc switch to select ABI level ( "-mabi=${CT_ARCH_ABI}" )
832 - CT_ARCH_CPU_CFLAG : the gcc switch to select CPU instruction set ( "-mcpu=${CT_ARCH_CPU}" )
833 - CT_ARCH_TUNE_CFLAG : the gcc switch to select scheduling ( "-mtune=${CT_ARCH_TUNE}" )
834 - CT_ARCH_FPU_CFLAG : the gcc switch to select FPU type ( "-mfpu=${CT_ARCH_FPU}" )
835 - CT_ARCH_FLOAT_CFLAG : the gcc switch to choose floating point arithmetics ( "-msoft-float" or /empty/ )
836 - CT_ARCH_ENDIAN_CFLAG : the gcc switch to choose big or little endian ( "-mbig-endian" or "-mlittle-endian" )
841 - the environement variables to configure the core and final compiler, specific to this architecture:
842 - CT_ARCH_CC_CORE_EXTRA_CONFIG : additional, architecture specific core gcc ./configure flags
843 - CT_ARCH_CC_EXTRA_CONFIG : additional, architecture specific final gcc ./configure flags
848 - the architecture-specific CFLAGS and LDFLAGS:
849 - CT_ARCH_TARGET_CLFAGS
850 - CT_ARCH_TARGET_LDFLAGS
854 You can have a look at "config/arch/arm.in" and "scripts/build/arch/arm.sh" for
855 a quite complete example of what an actual architecture description looks like.
860 A kernel is defined by:
862 - a human-readable name, in lower case letters, with numbers as appropriate.
863 The underscore is allowed; space and special characters are not (although
864 they are internally replaced with underscores.
865 Eg.: linux, bare-metal
866 - a file in "config/kernel/", named after the kernel name, and suffixed with
868 Eg.: config/kernel/linux.in, config/kernel/bare-metal.in
869 - a file in "scripts/build/kernel/", named after the kernel name, and suffixed
871 Eg.: scripts/build/kernel/linux.sh, scripts/build/kernel/bare-metal.sh
873 The kernel's ".in" file must contain:
874 > an optional lines containing exactly "# EXPERIMENTAL", starting on the
875 first column, and without any following space or other character.
876 If this line is present, then this kernel is considered EXPERIMENTAL,
877 and correct dependency on EXPERIMENTAL will be set.
879 > the config option "KERNEL_%kernel_name%" (where %kernel_name% is to be
880 replaced with the actual kernel name, with all special characters and
881 spaces replaced by underscores).
882 That config option must have *neither* a type, *nor* a prompt! Also, it can
883 *not* depends on EXPERIMENTAL.
884 Eg.: KERNEL_linux, KERNEL_bare_metal
886 defines a (terse) help entry for this kernel.
888 config KERNEL_bare_metal
890 Build a compiler for use without any kernel.
892 selects adequate associated config options.
894 config KERNEL_bare_metal
897 Build a compiler for use without any kernel.
899 > other kernel specific options, at your discretion. Note however that, to
900 avoid name-clashing, such options should be prefixed with
901 "KERNEL_%kernel_name%", where %kernel_name% is again tp be replaced with
902 the actual kernel name.
903 (Note: due to historical reasons, and lack of time to clean up the code,
904 I may have left some config options that do not completely conform to
905 this, as the kernel name was written all upper case. However, the prefix
906 is unique among kernels, and does not cause harm).
908 The kernel's ".sh" file API:
909 > is a bash script fragment
911 > defines the function CT_DoKernelTupleValues
912 + see the architecture's CT_DoArchTupleValues, except for:
913 + set the environment variable CT_TARGET_KERNEL, the kernel part of the
915 + return value: ignored
917 > defines the function "do_kernel_get":
920 - all variables from the ".config" file.
921 + return value: 0 for success, !0 for failure.
922 + behavior: download the kernel's sources, and store the tarball into
923 "${CT_TARBALLS_DIR}". To this end, a functions is available, that
924 abstracts downloading tarballs:
925 - CT_DoGet <tarball_base_name> <URL1 [URL...]>
926 Eg.: CT_DoGet linux-2.6.26.5 ftp://ftp.kernel.org/pub/linux/kernel/v2.6
927 Note: retrieving sources from svn, cvs, git and the likes is not supported
928 by CT_DoGet. You'll have to do this by hand, as it is done for eglibc in
929 "scripts/build/libc/eglibc.sh"
931 > defines the function "do_kernel_extract":
934 - all variables from the ".config" file,
935 + return value: 0 for success, !0 for failure.
936 + behavior: extract the kernel's tarball into "${CT_SRC_DIR}", and apply
937 required patches. To this end, a function is available, that abstracts
939 - CT_ExtractAndPatch <tarball_base_name>
940 Eg.: CT_ExtractAndPatch linux-2.6.26.5
942 > defines the function "do_kernel_headers":
945 - all variables from the ".config" file,
946 + return value: 0 for success, !0 for failure.
947 + behavior: install the kernel headers (if any) in "${CT_SYSROOT_DIR}/usr/include"
949 > defines any kernel-specific helper functions
950 These functions, if any, must be prefixed with "do_kernel_%CT_KERNEL%_",
951 where '%CT_KERNEL%' is to be replaced with the actual kernel name, to avoid
954 You can have a look at "config/kernel/linux.in" and "scripts/build/kernel/linux.sh"
955 as an example of what a complex kernel description looks like.
957 Adding a new version of a component |
958 ------------------------------------+
960 When a new component, such as the Linux kernel, gcc or any other is released,
961 adding the new version to crosstool-NG is quite easy. There is a script that
962 will do all that for you:
963 scripts/addToolVersion.sh
965 Run it with no option to get some help.
970 To Be Written later...