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
48 crosstool-NG aims at building toolchains. Toolchains are an essential component
49 in a software development project. It will compile, assemble and link the code
50 that is being developed. Some pieces of the toolchain will eventually end up
51 in the resulting binary/ies: static libraries are but an example.
53 So, a toolchain is a very sensitive piece of software, as any bug in one of the
54 components, or a poorly configured component, can lead to execution problems,
55 ranging from poor performance, to applications ending unexpectedly, to
56 mis-behaving software (which more than often is hard to detect), to hardware
57 damage, or even to human risks (which is more than regrettable).
59 Toolchains are made of different piece of software, each being quite complex
60 and requiring specially crafted options to build and work seamlessly. This
61 is usually not that easy, even in the not-so-trivial case of native toolchains.
62 The work reaches a higher degree of complexity when it comes to cross-
63 compilation, where it can become quite a nightmare...
65 Some cross-toolchains exist on the internet, and can be used for general
66 development, but they have a number of limitations:
67 - they can be general purpose, in that they are configured for the majority:
68 no optimisation for your specific target,
69 - they can be prepared for a specific target and thus are not easy to use,
70 nor optimised for, or even supporting your target,
71 - they often are using aging components (compiler, C library, etc...) not
72 supporting special features of your shiny new processor;
73 On the other side, these toolchain offer some advantages:
74 - they are ready to use and quite easy to install and setup,
75 - they are proven if used by a wide community.
77 But once you want to get all the juice out of your specific hardware, you will
78 want to build your own toolchain. This is where crosstool-NG comes into play.
80 There are also a number of tools that build toolchains for specific needs,
81 which are not really scalable. Examples are:
82 - buildroot (buildroot.uclibc.org) whose main purpose is to build root file
83 systems, hence the name. But once you have your toolchain with buildroot,
84 part of it is installed in the root-to-be, so if you want to build a whole
85 new root, you either have to save the existing one as a template and
86 restore it later, or restart again from scratch. This is not convenient,
87 - ptxdist (www.pengutronix.de/software/ptxdist), whose purpose is very
89 - other projects (openembedded.org for example), which are again used to
90 build root file systems.
92 crosstool-NG is really targeted at building toolchains, and only toolchains.
93 It is then up to you to use it the way you want.
100 crosstool was first 'conceived' by Dan Kegel, who offered it to the community
101 as a set of scripts, a repository of patches, and some pre-configured, general
102 purpose setup files to be used to configure crosstool. This is available at
103 http://www.kegel.com/crosstool, and the subversion repository is hosted on
104 google at http://code.google.com/p/crosstool/.
106 I once managed to add support for uClibc-based toolchains, but it did not make
107 into mainline, mostly because I didn't have time to port the patch forward to
108 the new versions, due in part to the big effort it was taking.
110 So I decided to clean up crosstool in the state it was, re-order the things
111 in place, add appropriate support for what I needed, that is uClibc support
112 and a menu-driven configuration, named the new implementation crosstool-NG,
113 (standing for crosstool Next Generation, as many other comunity projects do,
114 and as a wink at the TV series "Star Trek: The Next Generation" ;-) ) and
115 made it available to the community, in case it was of interest to any one.
117 ___________________________
119 Installing crosstool-NG /
120 ________________________/
122 There are two ways you can use crosstool-NG:
123 - build and install it, then get rid of the sources like you'd do for most
125 - or only build it and run from the source directory.
127 The former should be used if you got crosstool-NG from a packaged tarball, see
128 "Install method", below, while the latter is most useful for developpers that
129 use a clone of the repository, and want to submit patches, see "The Hacker's
135 If you go for the install, then you just follow the classical, but yet easy
137 ./configure --prefix=/some/place
140 export PATH="${PATH}:/some/place/bin"
142 You can then get rid of crosstool-NG source. Next create a directory to serve
143 as a working place, cd in there and run:
146 See below for complete usage.
151 If you go the hacker's way, then the usage is a bit different, although very
156 Now, *do not* remove crosstool-NG sources. They are needed to run crosstool-NG!
157 Stay in the directory holding the sources, and run:
160 See below for complete usage.
162 Now, provided you used a clone of the repository, you can send me your changes.
163 See the section titled CONTRIBUTING, below, for how to submit changees.
165 Preparing for packaging |
166 ------------------------+
168 If you plan on packaging crosstool-NG, you surely don't want to install it
169 in your root file system. The install procedure of crosstool-NG honors the
172 ./configure --prefix=/usr
174 make DESTDIR=/packaging/place install
179 crosstool-NG comes with a shell script fragment that defines bash-compatible
180 completion. That shell fragment is currently not installed automatically, but
183 To install the shell script fragment, you have two options:
184 - install system-wide, most probably by copying ct-ng.comp into
185 /etc/bash_completion.d/
186 - install for a single user, by copying ct-ng.comp into ${HOME}/ and
187 sourcing this file from your ${HOME}/.bashrc
192 Some people contibuted code that couldn't get merged for various reasons. This
193 code is available as lzma-compressed patches, in the contrib/ sub-directory.
194 These patches are to be applied to the source of crosstool-NG, prior to
195 installing, using something like the following:
196 lzcat contrib/foobar.patch.lzma |patch -p1
198 There is no guarantee that a particuliar contribution applies to the current
199 version of crosstool-ng, or that it will work at all. Use contributions at
202 ____________________________
204 Configuring crosstool-NG /
205 _________________________/
207 crosstool-NG is configured with a configurator presenting a menu-stuctured set
208 of options. These options let you specify the way you want your toolchain
209 built, where you want it installed, what architecture and specific processor it
210 will support, the version of the components you want to use, etc... The
211 value for those options are then stored in a configuration file.
213 The configurator works the same way you configure your Linux kernel. It is
214 assumed you now how to handle this.
216 To enter the menu, type:
219 Almost every config item has a help entry. Read them carefully.
221 String and number options can refer to environment variables. In such a case,
222 you must use the shell syntax: ${VAR}. You shall neither single- nor double-
223 quote the string/number options.
225 There are three environment variables that are computed by crosstool-NG, and
229 It represents the target tuple you are building for. You can use it for
230 example in the installation/prefix directory, such as:
231 /opt/x-tools/${CT_TARGET}
234 The top directory where crosstool-NG is running. You shouldn't need it in
235 most cases. There is one case where you may need it: if you have local
236 patches and you store them in your running directory, you can refer to them
237 by using CT_TOP_DIR, such as:
238 ${CT_TOP_DIR}/patches.myproject
241 The version of crosstool-NG you are using. Not much use for you, but it's
242 there if you need it.
244 Interesting config options |
245 ---------------------------+
247 CT_LOCAL_TARBALLS_DIR:
248 If you already have some tarballs in a direcotry, enter it here. That will
249 speed up the retrieving phase, where crosstool-NG would otherwise download
253 This is where the toolchain will be installed in (and for now, where it
254 will run from). Common use is to add the target tuple in the directory
255 path, such as (see above):
256 /opt/x-tools/${CT_TARGET}
259 An identifier for your toolchain, will take place in the vendor part of the
260 target tuple. It shall *not* contain spaces or dashes. Usually, keep it
261 to a one-word string, or use underscores to separate words if you need.
262 Avoid dots, commas, and special characters.
265 An alias for the toolchian. It will be used as a prefix to the toolchain
266 tools. For example, you will have ${CT_TARGET_ALIAS}-gcc
268 Also, if you think you don't see enough versions, you can try to enable one of
272 Show obsolete versions or tools. Most of the time, you don't want to base
273 your toolchain on too old a version (of gcc, for example). But at times, it
274 can come handy to use such an old version for regression tests. Those old
275 versions are hidden behind CT_OBSOLETE. Those versions (or features) are so
276 marked because maintaining support for those in crosstool-NG would be too
277 costly, time-wise, and time is dear.
280 Show experimental versions or tools. Again, you might not want to base your
281 toolchain on too recent tools (eg. gcc) for production. But if you need a
282 feature present only in a recent version, or a new tool, you can find them
283 hidden behind CT_EXPERIMENTAL. Those versions (or features) did not (yet)
284 receive thorough testing in crosstool-NG, and/or are not mature enough to
287 Re-building an existing toolchain |
288 ----------------------------------+
290 If you have an existing toolchain, you can re-use the options used to build it
291 to create a new toolchain. That needs a very little bit of effort on your side
292 but is quite easy. The options to build a toolchain are saved with the
293 toolchain, and you can retrieve this configuration by running:
296 This will dump the configuration to stdout, so to rebuild a toolchain with this
297 configuration, the following is all you need to do:
298 ${CT_TARGET}-config >.config
301 Then, you can review and change the configuration by running:
304 ________________________
306 Running crosstool-NG /
307 _____________________/
309 To build the toolchain, simply type:
312 This will use the above configuration to retrieve, extract and patch the
313 components, build, install and eventually test your newly built toolchain.
315 You are then free to add the toolchain /bin directory in your PATH to use
318 In any case, you can get some terse help. Just type:
323 Stopping and restarting a build |
324 --------------------------------+
326 If you want to stop the build after a step you are debugging, you can pass the
327 variable STOP to make:
330 Conversely, if you want to restart a build at a specific step you are
331 debugging, you can pass the RESTART variable to make:
332 ct-ng RESTART=some_step
334 Alternatively, you can call make with the name of a step to just do that step:
337 ct-ng RESTART=libc_headers STOP=libc_headers
339 The shortcuts +step_name and step_name+ allow to respectively stop or restart
341 ct-ng +libc_headers and: ct-ng libc_headers+
343 ct-ng STOP=libc_headers and: ct-ng RESTART=libc_headers
345 To obtain the list of acceptable steps, please call:
348 Note that in order to restart a build, you'll have to say 'Y' to the config
349 option CT_DEBUG_CT_SAVE_STEPS, and that the previous build effectively went
352 Building all toolchains at once |
353 --------------------------------+
355 You can build all samples; simply call:
358 Overriding the number of // jobs |
359 ---------------------------------+
361 If you want to override the number of jobs to run in // (the -j option to
362 make), you can either re-enter the menuconfig, or simply add it on the command
366 which tells crosstool-NG to override the number of // jobs to 4.
368 You can see the actions that support overriding the number of // jobs in
369 the help menu. Those are the ones with [.#] after them (eg. build[.#] or
370 build-all[.#], and so on...).
375 The crosstool-NG script 'ct-ng' is a Makefile-script. It does *not* execute
376 in parallel (there is not much to gain). When speaking of // jobs, we are
377 refering to the number of // jobs when making the *components*. That is, we
378 speak of the number of // jobs used to build gcc, glibc, and so on...
383 Starting with gcc-4.3 come two new dependencies: GMP and MPFR. With gcc-4.4,
384 come three new ones: GMP, PPL and CLooG/ppl. These are libraries that enable
385 advanced features to gcc. Additionally, some of the libraries can be used by
386 binutils and gdb. Unfortunately, not all systems on which crosstool-NG runs
387 have all of those libraries. And for those that do, the versions of those
388 libraries may be older than the version required by gcc.
390 This is why crosstool-NG builds its own set of libraries as part of the
393 The libraries are built as shared libraries, because building them as static
394 libraries has some short-comings. This poses no problem at build time, as
395 crosstool-NG correctly points gcc (and binutils and gdb) to the correct
396 place where our own version of the libraries are installed. But it poses
397 a problem when gcc et al. are run: the place where the libraries are is most
398 probably not known to the host dynamic linker. Still worse, if the host system
399 has its own versions, then ld.so would load the wrong library!
401 So we have to force the dynamic linker to load the correct version. We do this
402 by using the LD_LIBRARY_PATH variable, that informs the dynamic linker where
403 to look for shared libraries prior to searching its standard places. But we
404 can't impose that burden on all the system (because it'd be a nightmare to
405 configure, and because two toolchains on the same system may use different
406 versions of the libraries); so we have to do it on a per-toolchain basis.
408 So we rename all binaries of the toolchain (by adding a dot '.' as their first
409 character), and add a small program, the so-called "tools wrapper", that
410 correctly sets LD_LIBRARY_PATH prior to running the real tool.
412 First, the wrapper was written as a POSIX-compliant shell script. That shell
413 script is very simple, if not trivial, and works great. The only drawback is
414 that it does not work on host systems that lack a shell, for example the
415 MingW32 environment. To solve the issue, the wrapper has been re-written in C,
416 and compiled at build time. This C wrapper is much more complex than the shell
417 script, and although it sems to be working, it's been only lightly tested.
418 Some of the expected short-comings with this C wrapper are;
419 - multi-byte file names may not be handled correctly
420 - it's really big for what it does
422 So, the default wrapper installed with your toolchain is the shell script.
423 If you know that your system is missing a shell, then you shall use the C
424 wrapper (and report back whether it works, or does not work, for you).
427 _______________________
429 Using the toolchain /
430 ____________________/
432 Using the toolchain is as simple as adding the toolchain's bin directory in
434 export PATH="${PATH}:/your/toolchain/path/bin"
436 and then using the target tuple to tell the build systems to use your
438 ./configure --target=your-target-tuple
440 make CC=your-target-tuple-gcc
442 make CROSS_COMPILE=your-target-tuple-
445 It is strongly advised not to use the toolchain sys-root directory as an
446 install directory for your programs/packages. If you do so, you will not be
447 able to use your toolchain for another project. It is even strongly advised
448 that your toolchain is chmod-ed to read-only once successfully build, so that
449 you don't go polluting your toolchain with your programs/packages' files.
451 Thus, when you build a program/package, install it in a separate directory,
452 eg. /your/root. This directory is the /image/ of what would be in the root file
453 system of your target, and will contain all that your programs/packages have
456 The 'populate' script |
457 ----------------------+
459 When your root directory is ready, it is still missing some important bits: the
460 toolchain's libraries. To populate your root directory with those libs, just
462 your-target-tuple-populate -s /your/root -d /your/root-populated
464 This will copy /your/root into /your/root-populated, and put the needed and only
465 the needed libraries there. Thus you don't polute /your/root with any cruft that
466 would no longer be needed should you have to remove stuff. /your/root always
467 contains only those things you install in it.
469 You can then use /your/root-populated to build up your file system image, a
470 tarball, or to NFS-mount it from your target, or whatever you need.
472 The populate script accepts the following options:
475 Use 'src_dir' as the un-populated root directory.
478 Put the populated root directory in 'dst_dir'.
481 Always add specified libraries.
484 Always add libraries listed in 'file'.
487 Remove 'dst_dir' if it previously existed; continue even if any library
488 specified with -l or -L is missing.
491 Be verbose, and tell what's going on (you can see exactly where libs are
497 See 'your-target-tuple-populate -h' for more information on the options.
499 Here is how populate works:
501 1) performs some sanity checks:
502 - src_dir and dst_dir are specified
504 - unless forced, dst_dir does not exist
507 2) copy src_dir to dst_dir
509 3) add forced libraries to dst_dir
510 - build the list from -l and -L options
511 - get forced libraries from the sysroot (see below for heuristics)
512 - abort on the first missing library, unless -f is specified
514 4) add all missing libraries to dst_dir
515 - scan dst_dir for every ELF files that are 'executable' or
517 - list the "NEEDED Shared library" fields
518 - check if the library is already in dst_dir/lib or dst_dir/usr/lib
519 - if not, get the library from the sysroot
520 - if it's in sysroot/lib, copy it to dst_dir/lib
521 - if it's in sysroot/usr/lib, copy it to dst_dir/usr/lib
522 - in both cases, use the SONAME of the library to create the file
524 - if it was not found in the sysroot, this is an error.
531 There are four kinds of toolchains you could encounter.
533 First off, you must understand the following: when it comes to compilers there
534 are up to four machines involved:
535 1) the machine configuring the toolchain components: the config machine
536 2) the machine building the toolchain components: the build machine
537 3) the machine running the toolchain: the host machine
538 4) the machine the toolchain is generating code for: the target machine
540 We can most of the time assume that the config machine and the build machine
541 are the same. Most of the time, this will be true. The only time it isn't
542 is if you're using distributed compilation (such as distcc). Let's forget
543 this for the sake of simplicity.
545 So we're left with three machines:
550 Any toolchain will involve those three machines. You can be as pretty sure of
551 this as "2 and 2 are 4". Here is how they come into play:
553 1) build == host == target
554 This is a plain native toolchain, targetting the exact same machine as the
555 one it is built on, and running again on this exact same machine. You have
556 to build such a toolchain when you want to use an updated component, such
557 as a newer gcc for example.
558 crosstool-NG calls it "native".
560 2) build == host != target
561 This is a classic cross-toolchain, which is expected to be run on the same
562 machine it is compiled on, and generate code to run on a second machine,
564 crosstool-NG calls it "cross".
566 3) build != host == target
567 Such a toolchain is also a native toolchain, as it targets the same machine
568 as it runs on. But it is build on another machine. You want such a
569 toolchain when porting to a new architecture, or if the build machine is
570 much faster than the host machine.
571 crosstool-NG calls it "cross-native".
573 4) build != host != target
574 This one is called a canadian-toolchain (*), and is tricky. The three
575 machines in play are different. You might want such a toolchain if you
576 have a fast build machine, but the users will use it on another machine,
577 and will produce code to run on a third machine.
578 crosstool-NG calls it "canadian".
580 crosstool-NG can build all these kinds of toolchains (or is aiming at it,
583 (*) The term Canadian Cross came about because at the time that these issues
584 were all being hashed out, Canada had three national political parties.
585 http://en.wikipedia.org/wiki/Cross_compiler
587 Seemingly-native toolchains |
588 ----------------------------+
590 Seemingly-native toolchains are toolchains that target the same architecture
591 as the one it is built on, and on which it will run, but the machine tuple
592 may be different (eg i686 vs. i386, or x86_64-unknown-linux-gnu vs.
593 x86_64-pc-linux-gnu). This also applies if the target architecture is of the
594 same kind (eg. x86 vs. x86_64, or ppc vs. ppc64).
596 Such toolchain is tricky to build, as the configure scripts may incorrectly
597 assume that files (headers and libs) from the build (or host) machine can be
598 used by the cross-compiler it is going to build. The problem seems to arise
599 only with glibc (and eglibc?) starting with version 2.7.
607 Sending a bug report |
608 ---------------------+
610 If you need to send a bug report, please send a mail with subject
611 prefixed with "[CT_NG]" with to following destinations:
612 TO: yann.morin.1998 (at) anciens.enib.fr
613 CC: crossgcc (at) sourceware.org
618 If you want to enhance crosstool-NG, there's a to-do list in the TODO file.
620 Patches should come with the appropriate SoB line. A SoB line is typically
622 Signed-off-by: John DOE <john.doe@somewhere.net>
624 The SoB line is clearly described in Documentation/SubmittingPatches , section
625 12, of your favourite Linux kernel source tree.
627 Then you'll need to correctly configure Mercurial. There are two extensions
628 that you may find usefull:
629 - mq : http://mercurial.selenic.com/wiki/MqExtension
630 - patchbomb : http://mercurial.selenic.com/wiki/PatchbombExtension
632 Commit messages should look like (without leading pipes):
633 |component: short, one-line description
635 |optional longer description
636 |on multiple lines if needed
638 Here is an example commit message (see revision a53a5e1d61db):
639 |comp-libs/cloog: fix building
641 |For CLooG/PPL 0.15.3, the directory name was simply cloog-ppl.
642 |For any later versions, the directory name does have the version, such as
645 Here's a typical hacking session:
646 hg clone http://ymorin.is-a-geek.org/hg/crosstool-ng crosstool-ng
649 hg qnew -D -U -e my_first_patch
650 *edit patch description*
651 *hack* *hack* *check* *fails* *hack* *hack* *check* *works*
653 *edit patch description, serving as commit message*
654 hg qnew -D -U -e my_second_patch
655 *edit patch description*
656 *hack* *hack* *check* *fails* *hack* *hack* *check* *works*
658 *edit patch description, serving as commit message*
659 hg email --outgoing --intro \
660 --from '"Your Full NAME" <your.email (at) your.domain>' \
661 --to '"Yann E. MORIN" <yann.morin.1998 (at) anciens.enib.fr>' \
662 --cc 'crossgcc (at) sourceware.org'
663 *edit introductory message*
665 *re-send if no answer for a few days*
667 Note: replace '(at)' above with a plain '@'.
675 Internally, crosstool-NG is script-based. To ease usage, the frontend is
681 The entry point to crosstool-NG is the Makefile script "ct-ng". Calling this
682 script with an action will act exactly as if the Makefile was in the current
683 working directory and make was called with the action as rule. Thus:
686 is equivalent to having the Makefile in CWD, and calling:
689 Having ct-ng as it is avoids copying the Makefile everywhere, and acts as a
692 ct-ng loads sub- Makefiles from the library directory $(CT_LIB_DIR), as set up
693 at configuration time with ./configure.
695 ct-ng also searches for config files, sub-tools, samples, scripts and patches in
696 that library directory.
698 Because of a stupid make behavior/bug I was unable to track down, implicit make
699 rules are disabled: installing with --local would triger those rules, and mconf
705 The kconfig language is a hacked version, vampirised from the Linux kernel
706 (http://www.kernel.org/), and (heavily) adapted to my needs.
708 The list of the most notable changes (at least the ones I remember) follows:
709 - the CONFIG_ prefix has been replaced with CT_
710 - a leading | in prompts is skipped, and subsequent leading spaces are not
712 - otherwise leading spaces are silently trimmed
714 The kconfig parsers (conf and mconf) are not installed pre-built, but as
715 source files. Thus you can have the directory where crosstool-NG is installed,
716 exported (via NFS or whatever) and have clients with different architectures
717 use the same crosstool-NG installation, and most notably, the same set of
720 Architecture-specific |
721 ----------------------+
723 Note: this chapter is not really well written, and might thus be a little bit
724 complex to understand. To get a better grasp of what an architecture is, the
725 reader is kindly encouraged to look at the "arch/" sub-directory, and to the
726 existing architectures to see how things are laid out.
728 An architecture is defined by:
730 - a human-readable name, in lower case letters, with numbers as appropriate.
731 The underscore is allowed; space and special characters are not.
733 - a file in "config/arch/", named after the architecture's name, and suffixed
735 Eg.: config/arch/arm.in
736 - a file in "scripts/build/arch/", named after the architecture's name, and
738 Eg.: scripts/build/arch/arm.sh
740 The architecture's ".in" file API:
741 > the config option "ARCH_%arch%" (where %arch% is to be replaced with the
742 actual architecture name).
743 That config option must have *neither* a type, *nor* a prompt! Also, it can
744 *not* depend on any other config option (EXPERIMENTAL is managed as above).
748 defines a (terse) help entry for this architecture:
752 The ARM architecture.
754 selects adequate associated config options.
755 Note: 64-bit architectures *shall* select ARCH_64
758 select ARCH_SUPPORTS_BOTH_ENDIAN
759 select ARCH_DEFAULT_LE
761 The ARM architecture.
766 The x86_64 architecture.
768 > other target-specific options, at your discretion. Note however that to
769 avoid name-clashing, such options shall be prefixed with "ARCH_%arch%",
770 where %arch% is again replaced by the actual architecture name.
771 (Note: due to historical reasons, and lack of time to clean up the code,
772 I may have left some config options that do not completely conform to
773 this, as the architecture name was written all upper case. However, the
774 prefix is unique among architectures, and does not cause harm).
776 The architecture's ".sh" file API:
777 > the function "CT_DoArchTupleValues"
780 - all variables from the ".config" file,
781 - the two variables "target_endian_eb" and "target_endian_el" which are
782 the endianness suffixes
783 + return value: 0 upon success, !0 upon failure
786 - the environment variable CT_TARGET_ARCH
788 the architecture part of the target tuple.
789 Eg.: "armeb" for big endian ARM
793 - the environment variable CT_TARGET_SYS
795 the sytem part of the target tuple.
796 Eg.: "gnu" for glibc on most architectures
797 "gnueabi" for glibc on an ARM EABI
799 - for glibc-based toolchain: "gnu"
800 - for uClibc-based toolchain: "uclibc"
803 - the environment variable CT_KERNEL_ARCH
805 the architecture name as understandable by the Linux kernel build
807 Eg.: "arm" for an ARM
808 "powerpc" for a PowerPC
814 - the environment variables to configure the cross-gcc (defaults)
815 - CT_ARCH_WITH_ARCH : the gcc ./configure switch to select architecture level ( "--with-arch=${CT_ARCH_ARCH}" )
816 - CT_ARCH_WITH_ABI : the gcc ./configure switch to select ABI level ( "--with-abi=${CT_ARCH_ABI}" )
817 - CT_ARCH_WITH_CPU : the gcc ./configure switch to select CPU instruction set ( "--with-cpu=${CT_ARCH_CPU}" )
818 - CT_ARCH_WITH_TUNE : the gcc ./configure switch to select scheduling ( "--with-tune=${CT_ARCH_TUNE}" )
819 - CT_ARCH_WITH_FPU : the gcc ./configure switch to select FPU type ( "--with-fpu=${CT_ARCH_FPU}" )
820 - CT_ARCH_WITH_FLOAT : the gcc ./configure switch to select floating point arithmetics ( "--with-float=soft" or /empty/ )
823 - the environment variables to pass to the cross-gcc to build target binaries (defaults)
824 - CT_ARCH_ARCH_CFLAG : the gcc switch to select architecture level ( "-march=${CT_ARCH_ARCH}" )
825 - CT_ARCH_ABI_CFLAG : the gcc switch to select ABI level ( "-mabi=${CT_ARCH_ABI}" )
826 - CT_ARCH_CPU_CFLAG : the gcc switch to select CPU instruction set ( "-mcpu=${CT_ARCH_CPU}" )
827 - CT_ARCH_TUNE_CFLAG : the gcc switch to select scheduling ( "-mtune=${CT_ARCH_TUNE}" )
828 - CT_ARCH_FPU_CFLAG : the gcc switch to select FPU type ( "-mfpu=${CT_ARCH_FPU}" )
829 - CT_ARCH_FLOAT_CFLAG : the gcc switch to choose floating point arithmetics ( "-msoft-float" or /empty/ )
830 - CT_ARCH_ENDIAN_CFLAG : the gcc switch to choose big or little endian ( "-mbig-endian" or "-mlittle-endian" )
835 - the environement variables to configure the core and final compiler, specific to this architecture:
836 - CT_ARCH_CC_CORE_EXTRA_CONFIG : additional, architecture specific core gcc ./configure flags
837 - CT_ARCH_CC_EXTRA_CONFIG : additional, architecture specific final gcc ./configure flags
842 - the architecture-specific CFLAGS and LDFLAGS:
843 - CT_ARCH_TARGET_CLFAGS
844 - CT_ARCH_TARGET_LDFLAGS
848 You can have a look at "config/arch/arm.in" and "scripts/build/arch/arm.sh" for
849 a quite complete example of what an actual architecture description looks like.
854 A kernel is defined by:
856 - a human-readable name, in lower case letters, with numbers as appropriate.
857 The underscore is allowed; space and special characters are not (although
858 they are internally replaced with underscores.
859 Eg.: linux, bare-metal
860 - a file in "config/kernel/", named after the kernel name, and suffixed with
862 Eg.: config/kernel/linux.in, config/kernel/bare-metal.in
863 - a file in "scripts/build/kernel/", named after the kernel name, and suffixed
865 Eg.: scripts/build/kernel/linux.sh, scripts/build/kernel/bare-metal.sh
867 The kernel's ".in" file must contain:
868 > an optional lines containing exactly "# EXPERIMENTAL", starting on the
869 first column, and without any following space or other character.
870 If this line is present, then this kernel is considered EXPERIMENTAL,
871 and correct dependency on EXPERIMENTAL will be set.
873 > the config option "KERNEL_%kernel_name%" (where %kernel_name% is to be
874 replaced with the actual kernel name, with all special characters and
875 spaces replaced by underscores).
876 That config option must have *neither* a type, *nor* a prompt! Also, it can
877 *not* depends on EXPERIMENTAL.
878 Eg.: KERNEL_linux, KERNEL_bare_metal
880 defines a (terse) help entry for this kernel.
882 config KERNEL_bare_metal
884 Build a compiler for use without any kernel.
886 selects adequate associated config options.
888 config KERNEL_bare_metal
891 Build a compiler for use without any kernel.
893 > other kernel specific options, at your discretion. Note however that, to
894 avoid name-clashing, such options should be prefixed with
895 "KERNEL_%kernel_name%", where %kernel_name% is again tp be replaced with
896 the actual kernel name.
897 (Note: due to historical reasons, and lack of time to clean up the code,
898 I may have left some config options that do not completely conform to
899 this, as the kernel name was written all upper case. However, the prefix
900 is unique among kernels, and does not cause harm).
902 The kernel's ".sh" file API:
903 > is a bash script fragment
905 > defines the function CT_DoKernelTupleValues
906 + see the architecture's CT_DoArchTupleValues, except for:
907 + set the environment variable CT_TARGET_KERNEL, the kernel part of the
909 + return value: ignored
911 > defines the function "do_kernel_get":
914 - all variables from the ".config" file.
915 + return value: 0 for success, !0 for failure.
916 + behavior: download the kernel's sources, and store the tarball into
917 "${CT_TARBALLS_DIR}". To this end, a functions is available, that
918 abstracts downloading tarballs:
919 - CT_DoGet <tarball_base_name> <URL1 [URL...]>
920 Eg.: CT_DoGet linux-2.6.26.5 ftp://ftp.kernel.org/pub/linux/kernel/v2.6
921 Note: retrieving sources from svn, cvs, git and the likes is not supported
922 by CT_DoGet. You'll have to do this by hand, as it is done for eglibc in
923 "scripts/build/libc/eglibc.sh"
925 > defines the function "do_kernel_extract":
928 - all variables from the ".config" file,
929 + return value: 0 for success, !0 for failure.
930 + behavior: extract the kernel's tarball into "${CT_SRC_DIR}", and apply
931 required patches. To this end, a function is available, that abstracts
933 - CT_ExtractAndPatch <tarball_base_name>
934 Eg.: CT_ExtractAndPatch linux-2.6.26.5
936 > defines the function "do_kernel_headers":
939 - all variables from the ".config" file,
940 + return value: 0 for success, !0 for failure.
941 + behavior: install the kernel headers (if any) in "${CT_SYSROOT_DIR}/usr/include"
943 > defines any kernel-specific helper functions
944 These functions, if any, must be prefixed with "do_kernel_%CT_KERNEL%_",
945 where '%CT_KERNEL%' is to be replaced with the actual kernel name, to avoid
948 You can have a look at "config/kernel/linux.in" and "scripts/build/kernel/linux.sh"
949 as an example of what a complex kernel description looks like.
951 Adding a new version of a component |
952 ------------------------------------+
954 When a new component, such as the Linux kernel, gcc or any other is released,
955 adding the new version to crosstool-NG is quite easy. There is a script that
956 will do all that for you:
957 scripts/addToolVersion.sh
959 Run it with no option to get some help.
964 To Be Written later...