docs/overview.txt
author "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Mon Oct 27 18:42:26 2008 +0000 (2008-10-27)
changeset 1001 c8ac48ba1411
parent 914 0b164a321177
child 1025 fc64f85e41cc
permissions -rw-r--r--
Enhance the make fragments:
- comment the different parts
- re-order the code so that it is homogeneous amogst fragments
- eye-candy in some existing comments

/trunk/tools/tools.mk | 17 15 2 0 +++++++++++++++--
/trunk/steps.mk | 38 26 12 0 ++++++++++++++++++++++++++------------
/trunk/samples/samples.mk | 41 28 13 0 ++++++++++++++++++++++++++++-------------
/trunk/config/config.mk | 2 1 1 0 +-
4 files changed, 70 insertions(+), 28 deletions(-)
yann@1
     1
File.........: overview.txt
yann@197
     2
Content......: Overview of how crosstool-NG works.
yann@92
     3
Copyrigth....: (C) 2007 Yann E. MORIN <yann.morin.1998@anciens.enib.fr>
yann@192
     4
License......: Creative Commons Attribution Share Alike (CC-by-sa), v2.5
yann@92
     5
yann@628
     6
____________________
yann@628
     7
                   /
yann@628
     8
Table Of Content  /
yann@628
     9
_________________/
yann@628
    10
yann@628
    11
yann@628
    12
Introduction
yann@628
    13
History
yann@628
    14
Installing crosstool-NG
yann@628
    15
  Install method
yann@628
    16
  The hacker's way
yann@837
    17
  Shell completion
yann@628
    18
  Contributed code
yann@628
    19
Configuring crosstool-NG
yann@628
    20
  Interesting config options
yann@628
    21
  Re-building an existing toolchain
yann@628
    22
Running crosstool-NG
yann@628
    23
  Stopping and restarting a build
yann@628
    24
  Testing all toolchains at once
yann@628
    25
  Overriding the number of // jobs
yann@628
    26
Using the toolchain
yann@628
    27
Toolchain types
yann@628
    28
Internals
yann@628
    29
  Makefile front-end
yann@628
    30
  Kconfig parser
yann@628
    31
  Architecture-specific
yann@628
    32
  Adding a new version of a component
yann@628
    33
  Build scripts
yann@628
    34
yann@1
    35
________________
yann@1
    36
               /
yann@1
    37
Introduction  /
yann@1
    38
_____________/
yann@1
    39
yann@1
    40
crosstool-NG aims at building toolchains. Toolchains are an essential component
yann@1
    41
in a software development project. It will compile, assemble and link the code
rpjday@436
    42
that is being developed. Some pieces of the toolchain will eventually end up
yann@1
    43
in the resulting binary/ies: static libraries are but an example.
yann@1
    44
yann@1
    45
So, a toolchain is a very sensitive piece of software, as any bug in one of the
yann@1
    46
components, or a poorly configured component, can lead to execution problems,
yann@1
    47
ranging from poor performance, to applications ending unexpectedly, to
yann@1
    48
mis-behaving software (which more than often is hard to detect), to hardware
rpjday@436
    49
damage, or even to human risks (which is more than regrettable).
yann@1
    50
yann@1
    51
Toolchains are made of different piece of software, each being quite complex
yann@1
    52
and requiring specially crafted options to build and work seamlessly. This
yann@1
    53
is usually not that easy, even in the not-so-trivial case of native toolchains.
yann@1
    54
The work reaches a higher degree of complexity when it comes to cross-
yann@40
    55
compilation, where it can become quite a nightmare...
yann@1
    56
yann@40
    57
Some cross-toolchains exist on the internet, and can be used for general
yann@1
    58
development, but they have a number of limitations:
yann@1
    59
  - they can be general purpose, in that they are configured for the majority:
yann@1
    60
    no optimisation for your specific target,
yann@1
    61
  - they can be prepared for a specific target and thus are not easy to use,
yann@1
    62
    nor optimised for, or even supporting your target,
rpjday@436
    63
  - they often are using aging components (compiler, C library, etc...) not
yann@1
    64
    supporting special features of your shiny new processor;
yann@1
    65
On the other side, these toolchain offer some advantages:
yann@1
    66
  - they are ready to use and quite easy to install and setup,
yann@1
    67
  - they are proven if used by a wide community.
yann@1
    68
yann@1
    69
But once you want to get all the juice out of your specific hardware, you will
yann@197
    70
want to build your own toolchain. This is where crosstool-NG comes into play.
yann@1
    71
rpjday@436
    72
There are also a number of tools that build toolchains for specific needs,
rpjday@436
    73
which are not really scalable. Examples are:
rpjday@436
    74
  - buildroot (buildroot.uclibc.org) whose main purpose is to build root file
yann@1
    75
    systems, hence the name. But once you have your toolchain with buildroot,
yann@1
    76
    part of it is installed in the root-to-be, so if you want to build a whole
yann@1
    77
    new root, you either have to save the existing one as a template and
yann@1
    78
    restore it later, or restart again from scratch. This is not convenient,
yann@1
    79
  - ptxdist (www.pengutronix.de/software/ptxdist), whose purpose is very
yann@1
    80
    similar to buildroot,
yann@702
    81
  - other projects (openembedded.org for example), which are again used to
yann@1
    82
    build root file systems.
yann@1
    83
rpjday@436
    84
crosstool-NG is really targeted at building toolchains, and only toolchains.
yann@1
    85
It is then up to you to use it the way you want.
yann@1
    86
yann@1
    87
___________
yann@1
    88
          /
yann@1
    89
History  /
yann@1
    90
________/
yann@1
    91
rpjday@436
    92
crosstool was first 'conceived' by Dan Kegel, who offered it to the community
yann@1
    93
as a set of scripts, a repository of patches, and some pre-configured, general
yann@1
    94
purpose setup files to be used to configure crosstool. This is available at
yann@203
    95
http://www.kegel.com/crosstool, and the subversion repository is hosted on
yann@203
    96
google at http://code.google.com/p/crosstool/.
yann@1
    97
yann@1
    98
I once managed to add support for uClibc-based toolchains, but it did not make
yann@437
    99
into mainline, mostly because I didn't have time to port the patch forward to
yann@1
   100
the new versions, due in part to the big effort it was taking.
yann@1
   101
yann@1
   102
So I decided to clean up crosstool in the state it was, re-order the things
yann@437
   103
in place, add appropriate support for what I needed, that is uClibc support
yann@437
   104
and a menu-driven configuration, named the new implementation crosstool-NG,
yann@437
   105
(standing for crosstool Next Generation, as many other comunity projects do,
yann@437
   106
and as a wink at the TV series "Star Trek: The Next Generation" ;-) ) and
yann@437
   107
made it available to the community, in case it was of interest to any one.
yann@1
   108
yann@294
   109
___________________________
yann@294
   110
                          /
yann@294
   111
Installing crosstool-NG  /
yann@294
   112
________________________/
yann@294
   113
yann@294
   114
There are two ways you can use crosstool-NG:
yann@294
   115
 - build and install it, then get rid of the sources like you'd do for most
yann@294
   116
   programs,
yann@294
   117
 - or only build it and run from the source directory.
yann@294
   118
yann@294
   119
The former should be used if you got crosstool-NG from a packaged tarball, see
rpjday@436
   120
"Install method", below, while the latter is most useful for developpers that
yann@294
   121
checked the code out from SVN, and want to submit patches, see "The Hacker's
yann@294
   122
way", below.
yann@294
   123
yann@294
   124
Install method |
yann@294
   125
---------------+
yann@294
   126
yann@294
   127
If you go for the install, then you just follow the classical, but yet easy
yann@294
   128
./configure way:
yann@294
   129
  ./configure --prefix=/some/place
yann@294
   130
  make
yann@294
   131
  make install
yann@294
   132
  export PATH="${PATH}:/some/place/bin"
yann@294
   133
yann@294
   134
You can then get rid of crosstool-NG source. Next create a directory to serve
yann@294
   135
as a working place, cd in there and run:
yann@294
   136
  ct-ng help
yann@294
   137
yann@294
   138
See below for complete usage.
yann@294
   139
yann@294
   140
The Hacker's way |
yann@294
   141
-----------------+
yann@294
   142
yann@294
   143
If you go the hacker's way, then the usage is a bit different, although very
yann@294
   144
simple:
yann@294
   145
  ./configure --local
yann@294
   146
  make
yann@294
   147
yann@294
   148
Now, *do not* remove crosstool-NG sources. They are needed to run crosstool-NG!
yann@294
   149
Stay in the directory holding the sources, and run:
yann@294
   150
  ./ct-ng help
yann@294
   151
yann@294
   152
See below for complete usage.
yann@294
   153
yann@294
   154
Now, provided you checked-out the code, you can send me your interesting changes
yann@294
   155
by running:
yann@294
   156
  svn diff
yann@294
   157
yann@294
   158
and mailing me the result! :-P
yann@294
   159
yann@837
   160
Shell completion |
yann@837
   161
-----------------+
yann@837
   162
yann@837
   163
crosstool-NG comes with a shell script fragment that defines bash-compatible
yann@837
   164
completion. That shell fragment is currently not installed automatically, but
yann@837
   165
this is planned.
yann@837
   166
yann@837
   167
To install the shell script fragment, you have two options:
yann@837
   168
 - install system-wide, most probably by copying ct-ng.comp into
yann@837
   169
   /etc/bash_completion.d/
yann@837
   170
 - install for a single user, by copying ct-ng.comp into ${HOME}/ and
yann@837
   171
   sourcing this file from your ${HOME}/.bashrc
yann@837
   172
yann@456
   173
Contributed code |
yann@456
   174
-----------------+
yann@456
   175
yann@456
   176
Some people contibuted code that couldn't get merged for various reasons. This
yann@456
   177
code is available as patches in the contrib/ sub-directory. These patches are
yann@456
   178
to be applied to the source of crosstool-NG, prior to installing.
yann@456
   179
yann@620
   180
An easy way to use contributed code is to pass the --with-contrib= option to
yann@620
   181
./configure. The possible values depend upon which contributions are packaged
yann@620
   182
with your version, but you can get with it with passing one of those two
yann@620
   183
special values:
yann@620
   184
  --with-contrib=list
yann@620
   185
    will list all available contributions
yann@620
   186
yann@620
   187
  --with-contrib=all
yann@620
   188
    will select all avalaible contributions
yann@620
   189
yann@620
   190
There is no guarantee that a particuliar contribution applies to the current
yann@620
   191
version of crosstool-ng, or that it will work at all. Use contributions at
yann@620
   192
your own risk.
yann@620
   193
yann@168
   194
____________________________
yann@168
   195
                           /
yann@168
   196
Configuring crosstool-NG  /
yann@168
   197
_________________________/
yann@168
   198
yann@620
   199
crosstool-NG is configured with a configurator presenting a menu-stuctured set
yann@620
   200
of options. These options let you specify the way you want your toolchain
yann@620
   201
built, where you want it installed, what architecture and specific processor it
yann@277
   202
will support, the version of the components you want to use, etc... The
yann@277
   203
value for those options are then stored in a configuration file.
yann@277
   204
yann@476
   205
The configurator works the same way you configure your Linux kernel. It is
yann@277
   206
assumed you now how to handle this.
yann@168
   207
yann@168
   208
To enter the menu, type:
yann@192
   209
  ct-ng menuconfig
yann@168
   210
yann@203
   211
Almost every config item has a help entry. Read them carefully.
yann@168
   212
yann@168
   213
String and number options can refer to environment variables. In such a case,
yann@192
   214
you must use the shell syntax: ${VAR}. You shall neither single- nor double-
yann@294
   215
quote the string/number options.
yann@168
   216
yann@192
   217
There are three environment variables that are computed by crosstool-NG, and
yann@168
   218
that you can use:
yann@168
   219
yann@168
   220
CT_TARGET:
yann@335
   221
  It represents the target tuple you are building for. You can use it for
yann@168
   222
  example in the installation/prefix directory, such as:
yann@168
   223
    /opt/x-tools/${CT_TARGET}
yann@168
   224
yann@168
   225
CT_TOP_DIR:
yann@182
   226
  The top directory where crosstool-NG is running. You shouldn't need it in
yann@182
   227
  most cases. There is one case where you may need it: if you have local
yann@182
   228
  patches and you store them in your running directory, you can refer to them
yann@168
   229
  by using CT_TOP_DIR, such as:
yann@168
   230
    ${CT_TOP_DIR}/patches.myproject
yann@168
   231
yann@168
   232
CT_VERSION:
yann@192
   233
  The version of crosstool-NG you are using. Not much use for you, but it's
yann@168
   234
  there if you need it.
yann@168
   235
yann@168
   236
Interesting config options |
yann@476
   237
---------------------------+
yann@168
   238
yann@168
   239
CT_LOCAL_TARBALLS_DIR:
yann@277
   240
  If you already have some tarballs in a direcotry, enter it here. That will
yann@197
   241
  speed up the retrieving phase, where crosstool-NG would otherwise download
yann@168
   242
  those tarballs.
yann@168
   243
yann@168
   244
CT_PREFIX_DIR:
yann@168
   245
  This is where the toolchain will be installed in (and for now, where it
yann@437
   246
  will run from). Common use is to add the target tuple in the directory
yann@277
   247
  path, such as (see above):
yann@277
   248
    /opt/x-tools/${CT_TARGET}
yann@168
   249
yann@168
   250
CT_TARGET_VENDOR:
yann@168
   251
  An identifier for your toolchain, will take place in the vendor part of the
yann@335
   252
  target tuple. It shall *not* contain spaces or dashes. Usually, keep it
yann@168
   253
  to a one-word string, or use underscores to separate words if you need.
yann@168
   254
  Avoid dots, commas, and special characters.
yann@168
   255
yann@168
   256
CT_TARGET_ALIAS:
yann@168
   257
  An alias for the toolchian. It will be used as a prefix to the toolchain
yann@168
   258
  tools. For example, you will have ${CT_TARGET_ALIAS}-gcc
yann@168
   259
yann@246
   260
Also, if you think you don't see enough versions, you can try to enable one of
yann@246
   261
those:
yann@246
   262
yann@246
   263
CT_OBSOLETE:
yann@246
   264
  Show obsolete versions or tools. Most of the time, you don't want to base
yann@246
   265
  your toolchain on too old a version (of gcc, for example). But at times, it
yann@246
   266
  can come handy to use such an old version for regression tests. Those old
yann@294
   267
  versions are hidden behind CT_OBSOLETE.
yann@246
   268
yann@246
   269
CT_EXPERIMENTAL:
yann@246
   270
  Show experimental versions or tools. Again, you might not want to base your
yann@246
   271
  toolchain on too recent tools (eg. gcc) for production. But if you need a
yann@246
   272
  feature present only in a recent version, or a new tool, you can find them
yann@246
   273
  hidden behind CT_EXPERIMENTAL.
yann@246
   274
yann@276
   275
Re-building an existing toolchain |
yann@276
   276
----------------------------------+
yann@276
   277
yann@276
   278
If you have an existing toolchain, you can re-use the options used to build it
yann@276
   279
to create a new toolchain. That needs a very little bit of effort on your side
yann@894
   280
but is quite easy. The options to build a toolchain are saved with the
yann@894
   281
toolchain, and you can retrieve this configuration by running:
yann@894
   282
  ${CT_TARGET}-config
yann@276
   283
yann@894
   284
This will dump the configuration to stdout, so to rebuild a toolchain with this
yann@894
   285
configuration, the following is all you need to do:
yann@894
   286
  ${CT_TARGET}-config >.config
yann@276
   287
yann@894
   288
Then, you can review and change the configuration by running:
yann@894
   289
  ct-ng menuconfig
yann@276
   290
yann@168
   291
________________________
yann@168
   292
                       /
yann@168
   293
Running crosstool-NG  /
yann@168
   294
_____________________/
yann@1
   295
yann@168
   296
To build the toolchain, simply type:
yann@203
   297
  ct-ng build
yann@135
   298
yann@135
   299
This will use the above configuration to retrieve, extract and patch the
yann@135
   300
components, build, install and eventually test your newly built toolchain.
yann@1
   301
yann@1
   302
You are then free to add the toolchain /bin directory in your PATH to use
yann@1
   303
it at will.
yann@1
   304
yann@135
   305
In any case, you can get some terse help. Just type:
yann@192
   306
  ct-ng help
yann@203
   307
or:
yann@203
   308
  man 1 ct-ng
yann@135
   309
rpjday@436
   310
Stopping and restarting a build |
yann@476
   311
--------------------------------+
yann@135
   312
yann@135
   313
If you want to stop the build after a step you are debugging, you can pass the
yann@135
   314
variable STOP to make:
yann@192
   315
  ct-ng STOP=some_step
yann@135
   316
yann@135
   317
Conversely, if you want to restart a build at a specific step you are
yann@135
   318
debugging, you can pass the RESTART variable to make:
yann@192
   319
  ct-ng RESTART=some_step
yann@135
   320
yann@136
   321
Alternatively, you can call make with the name of a step to just do that step:
yann@192
   322
  ct-ng libc_headers
yann@136
   323
is equivalent to:
yann@620
   324
  ct-ng RESTART=libc_headers STOP=libc_headers
yann@136
   325
yann@304
   326
The shortcuts +step_name and step_name+ allow to respectively stop or restart
yann@136
   327
at that step. Thus:
yann@304
   328
  ct-ng +libc_headers        and:    ct-ng libc_headers+
yann@136
   329
are equivalent to:
yann@192
   330
  ct-ng STOP=libc_headers    and:    ct-ng RESTART=libc_headers
yann@136
   331
yann@181
   332
To obtain the list of acceptable steps, please call:
yann@544
   333
  ct-ng list-steps
yann@181
   334
yann@168
   335
Note that in order to restart a build, you'll have to say 'Y' to the config
yann@168
   336
option CT_DEBUG_CT_SAVE_STEPS, and that the previous build effectively went
yann@168
   337
that far.
yann@92
   338
yann@168
   339
Testing all toolchains at once |
yann@476
   340
-------------------------------+
yann@92
   341
yann@168
   342
You can test-build all samples; simply call:
yann@192
   343
  ct-ng regtest
yann@40
   344
yann@335
   345
Overriding the number of // jobs |
yann@476
   346
---------------------------------+
yann@335
   347
yann@335
   348
If you want to override the number of jobs to run in // (the -j option to
yann@335
   349
make), you can either re-enter the menuconfig, or simply add it on the command
yann@335
   350
line, as such:
yann@335
   351
  ct-ng build.4
yann@335
   352
yann@335
   353
which tells crosstool-NG to override the number of // jobs to 4.
yann@335
   354
yann@335
   355
You can see the actions that support overriding the number of // jobs in
yann@335
   356
the help menu. Those are the ones with [.#] after them (eg. build[.#] or
yann@335
   357
regtest[.#], and so on...).
yann@335
   358
yann@227
   359
_______________________
yann@227
   360
                      /
yann@227
   361
Using the toolchain  /
yann@227
   362
____________________/
yann@227
   363
yann@227
   364
Using the toolchain is as simple as adding the toolchain's bin directory in
yann@227
   365
your PATH, such as:
yann@227
   366
  export PATH="${PATH}:/your/toolchain/path/bin"
yann@227
   367
yann@335
   368
and then using the target tuple to tell the build systems to use your
yann@227
   369
toolchain:
yann@335
   370
  ./configure --target=your-target-tuple
yann@294
   371
or
yann@335
   372
  make CC=your-target-tuple-gcc
yann@294
   373
or
yann@335
   374
  make CROSS_COMPILE=your-target-tuple-
yann@294
   375
and so on...
yann@227
   376
yann@476
   377
It is strongly advised not to use the toolchain sys-root directory as an
yann@620
   378
install directory for your programs/packages. If you do so, you will not be
yann@476
   379
able to use your toolchain for another project. It is even strongly advised
yann@476
   380
that your toolchain is chmod-ed to read-only once successfully build, so that
yann@620
   381
you don't go polluting your toolchain with your programs/packages' files.
yann@476
   382
yann@476
   383
Thus, when you build a program/package, install it in a separate directory,
yann@476
   384
eg. /your/root. This directory is the /image/ of what would be in the root file
yann@620
   385
system of your target, and will contain all that your programs/packages have
yann@476
   386
installed.
yann@476
   387
yann@227
   388
When your root directory is ready, it is still missing some important bits: the
yann@227
   389
toolchain's libraries. To populate your root directory with those libs, just
yann@227
   390
run:
yann@335
   391
  your-target-tuple-populate -s /your/root -d /your/root-populated
yann@227
   392
yann@227
   393
This will copy /your/root into /your/root-populated, and put the needed and only
yann@227
   394
the needed libraries there. Thus you don't polute /your/root with any cruft that
yann@227
   395
would no longer be needed should you have to remove stuff. /your/root always
yann@227
   396
contains only those things you install in it.
yann@227
   397
yann@227
   398
You can then use /your/root-populated to build up your file system image, a
yann@227
   399
tarball, or to NFS-mount it from your target, or whatever you need.
yann@227
   400
yann@294
   401
populate accepts the following options:
yann@294
   402
yann@294
   403
 -s [src_dir]
yann@294
   404
    Use 'src_dir' as the 'source', un-populated root directory
yann@294
   405
yann@294
   406
 -d [dst_dir]
yann@294
   407
    Put the 'destination', populated root directory in 'dst_dir'
yann@294
   408
yann@294
   409
 -f
yann@294
   410
    Remove 'dst_dir' if it previously existed
yann@294
   411
yann@294
   412
 -v
yann@294
   413
    Be verbose, and tell what's going on (you can see exactly where libs are
yann@294
   414
    coming from).
yann@294
   415
yann@294
   416
 -h
yann@294
   417
    Print the help
yann@294
   418
yann@40
   419
___________________
yann@40
   420
                  /
yann@40
   421
Toolchain types  /
yann@40
   422
________________/
yann@40
   423
yann@40
   424
There are four kinds of toolchains you could encounter.
yann@40
   425
yann@40
   426
First off, you must understand the following: when it comes to compilers there
yann@40
   427
are up to four machines involved:
yann@40
   428
  1) the machine configuring the toolchain components: the config machine
yann@40
   429
  2) the machine building the toolchain components:    the build machine
yann@40
   430
  3) the machine running the toolchain:                the host machine
yann@203
   431
  4) the machine the toolchain is generating code for: the target machine
yann@40
   432
yann@40
   433
We can most of the time assume that the config machine and the build machine
yann@40
   434
are the same. Most of the time, this will be true. The only time it isn't
yann@40
   435
is if you're using distributed compilation (such as distcc). Let's forget
yann@40
   436
this for the sake of simplicity.
yann@40
   437
yann@40
   438
So we're left with three machines:
yann@40
   439
 - build
yann@40
   440
 - host
yann@40
   441
 - target
yann@40
   442
yann@40
   443
Any toolchain will involve those three machines. You can be as pretty sure of
yann@40
   444
this as "2 and 2 are 4". Here is how they come into play:
yann@40
   445
yann@40
   446
1) build == host == target
yann@40
   447
    This is a plain native toolchain, targetting the exact same machine as the
yann@40
   448
    one it is built on, and running again on this exact same machine. You have
yann@40
   449
    to build such a toolchain when you want to use an updated component, such
yann@40
   450
    as a newer gcc for example.
yann@197
   451
    crosstool-NG calls it "native".
yann@40
   452
yann@40
   453
2) build == host != target
yann@40
   454
    This is a classic cross-toolchain, which is expected to be run on the same
yann@40
   455
    machine it is compiled on, and generate code to run on a second machine,
yann@40
   456
    the target.
yann@197
   457
    crosstool-NG calls it "cross".
yann@40
   458
yann@40
   459
3) build != host == target
yann@40
   460
    Such a toolchain is also a native toolchain, as it targets the same machine
yann@40
   461
    as it runs on. But it is build on another machine. You want such a
yann@40
   462
    toolchain when porting to a new architecture, or if the build machine is
yann@40
   463
    much faster than the host machine.
yann@197
   464
    crosstool-NG calls it "cross-native".
yann@40
   465
yann@40
   466
4) build != host != target
yann@92
   467
    This one is called a canadian-toolchain (*), and is tricky. The three
yann@40
   468
    machines in play are different. You might want such a toolchain if you
yann@40
   469
    have a fast build machine, but the users will use it on another machine,
yann@40
   470
    and will produce code to run on a third machine.
yann@197
   471
    crosstool-NG calls it "canadian".
yann@40
   472
yann@197
   473
crosstool-NG can build all these kinds of toolchains (or is aiming at it,
yann@197
   474
anyway!)
yann@40
   475
yann@40
   476
(*) The term Canadian Cross came about because at the time that these issues
yann@40
   477
    were all being hashed out, Canada had three national political parties.
yann@40
   478
    http://en.wikipedia.org/wiki/Cross_compiler
yann@40
   479
yann@1
   480
_____________
yann@1
   481
            /
yann@1
   482
Internals  /
yann@1
   483
__________/
yann@1
   484
yann@92
   485
Internally, crosstool-NG is script-based. To ease usage, the frontend is
yann@92
   486
Makefile-based.
yann@92
   487
yann@92
   488
Makefile front-end |
yann@476
   489
-------------------+
yann@92
   490
yann@203
   491
The entry point to crosstool-NG is the Makefile script "ct-ng". Calling this
yann@203
   492
script with an action will act exactly as if the Makefile was in the current
yann@203
   493
working directory and make was called with the action as rule. Thus:
yann@203
   494
  ct-ng menuconfig
yann@294
   495
yann@203
   496
is equivalent to having the Makefile in CWD, and calling:
yann@203
   497
  make menuconfig
yann@203
   498
yann@203
   499
Having ct-ng as it is avoids copying the Makefile everywhere, and acts as a
yann@203
   500
traditional command.
yann@203
   501
yann@203
   502
ct-ng loads sub- Makefiles from the library directory $(CT_LIB_DIR), as set up
yann@203
   503
at configuration time with ./configure.
yann@203
   504
yann@437
   505
ct-ng also searches for config files, sub-tools, samples, scripts and patches in
yann@203
   506
that library directory.
yann@92
   507
yann@294
   508
Because of a stupid make behavior/bug I was unable to track down, implicit make
yann@294
   509
rules are disabled: installing with --local would triger those rules, and mconf
yann@294
   510
was unbuildable.
yann@294
   511
yann@182
   512
Kconfig parser |
yann@476
   513
---------------+
yann@92
   514
yann@965
   515
The kconfig language is a hacked version, vampirised from the Linux kernel
yann@965
   516
(http://www.kernel.org/), and (heavily) adapted to my needs.
yann@92
   517
yann@203
   518
The kconfig parsers (conf and mconf) are not installed pre-built, but as
yann@203
   519
source files. Thus you can have the directory where crosstool-NG is installed,
yann@203
   520
exported (via NFS or whatever) and have clients with different architectures
yann@203
   521
use the same crosstool-NG installation, and most notably, the same set of
yann@203
   522
patches.
yann@203
   523
yann@381
   524
Architecture-specific |
yann@476
   525
----------------------+
yann@381
   526
yann@628
   527
Note: this chapter is not really well written, and might thus be a little bit
yann@628
   528
complex to understand. To get a better grasp of what an architecture is, the
yann@628
   529
reader is kindly encouraged to look at the "arch/" sub-directory, and to the
yann@628
   530
existing architectures to see how things are laid out.
yann@628
   531
yann@381
   532
An architecture is defined by:
yann@381
   533
yann@381
   534
 - a human-readable name, in lower case letters, with numbers as appropriate.
yann@628
   535
   The underscore is allowed; space and special characters are not.
yann@628
   536
     Eg.: arm, x86_64
yann@903
   537
 - a file in "config/arch/", named after the architecture's name, and suffixed
yann@903
   538
   with ".in".
yann@903
   539
     Eg.: config/arch/arm.in
yann@903
   540
 - a file in "scripts/build/arch/", named after the architecture's name, and
yann@903
   541
   suffixed with ".sh".
yann@903
   542
     Eg.: scripts/build/arch/arm.sh
yann@628
   543
yann@903
   544
The architecture's ".in" file API:
yann@628
   545
 > the config option "ARCH_%arch%" (where %arch% is to be replaced with the
yann@628
   546
   actual architecture name).
yann@628
   547
   That config option must have *neither* a type, *nor* a prompt! Also, it can
yann@628
   548
   *not* depend on any other config option (EXPERIMENTAL is managed as above).
yann@628
   549
     Eg.:
yann@628
   550
       config ARCH_arm
yann@630
   551
   + mandatory:
yann@702
   552
       defines a (terse) help entry for this architecture:
yann@630
   553
       Eg.:
yann@630
   554
         config ARCH_arm
yann@630
   555
           help
yann@630
   556
             The ARM architecture.
yann@628
   557
   + optional:
yann@628
   558
       selects adequate associated config options.
yann@628
   559
       Eg.:
yann@628
   560
         config ARCH_arm
yann@628
   561
           select ARCH_SUPPORTS_BOTH_ENDIAN
yann@628
   562
           select ARCH_DEFAULT_LE
yann@630
   563
           help
yann@630
   564
             The ARM architecture.
yann@628
   565
yann@628
   566
 > other target-specific options, at your discretion. Note however that to
yann@628
   567
   avoid name-clashing, such options shall be prefixed with "ARCH_%arch%",
yann@628
   568
   where %arch% is again replaced by the actual architecture name.
yann@628
   569
   (Note: due to historical reasons, and lack of time to clean up the code,
yann@628
   570
    I may have left some config options that do not completely conform to
yann@628
   571
    this, as the architecture name was written all upper case. However, the
yann@628
   572
    prefix is unique among architectures, and does not cause harm).
yann@381
   573
yann@903
   574
The architecture's ".sh" file API:
yann@965
   575
 > the function "CT_DoArchTupleValues"
yann@381
   576
   + parameters: none
yann@381
   577
   + environment:
yann@901
   578
     - all variables from the ".config" file,
yann@901
   579
     - the two variables "target_endian_eb" and "target_endian_el" which are
yann@901
   580
       the endianness suffixes
yann@381
   581
   + return value: 0 upon success, !0 upon failure
yann@381
   582
   + provides:
yann@391
   583
     - mandatory
yann@383
   584
     - the environment variable CT_TARGET_ARCH
yann@389
   585
     - contains:
yann@389
   586
       the architecture part of the target tuple.
yann@389
   587
       Eg.: "armeb" for big endian ARM
yann@389
   588
            "i386" for an i386
yann@389
   589
   + provides:
yann@391
   590
     - optional
yann@389
   591
     - the environment variable CT_TARGET_SYS
yann@456
   592
     - contains:
yann@383
   593
       the sytem part of the target tuple.
yann@383
   594
       Eg.: "gnu" for glibc on most architectures
yann@383
   595
            "gnueabi" for glibc on an ARM EABI
yann@383
   596
     - defaults to:
yann@383
   597
       - for glibc-based toolchain: "gnu"
yann@383
   598
       - for uClibc-based toolchain: "uclibc"
yann@383
   599
   + provides:
yann@383
   600
     - optional
yann@391
   601
     - the environment variable CT_KERNEL_ARCH
yann@383
   602
     - contains:
yann@391
   603
       the architecture name as understandable by the Linux kernel build
yann@391
   604
       system.
yann@391
   605
       Eg.: "arm" for an ARM
yann@391
   606
            "powerpc" for a PowerPC
yann@391
   607
            "i386" for an x86
yann@383
   608
     - defaults to:
yann@391
   609
       ${CT_ARCH}
yann@391
   610
   + provides:
yann@391
   611
     - optional
yann@767
   612
     - the environment variables to configure the cross-gcc (defaults)
yann@767
   613
       - CT_ARCH_WITH_ARCH    : the gcc ./configure switch to select architecture level         ( "--with-arch=${CT_ARCH_ARCH}"   )
yann@767
   614
       - CT_ARCH_WITH_ABI     : the gcc ./configure switch to select ABI level                  ( "--with-abi=${CT_ARCH_ABI}"     )
yann@767
   615
       - CT_ARCH_WITH_CPU     : the gcc ./configure switch to select CPU instruction set        ( "--with-cpu=${CT_ARCH_CPU}"     )
yann@767
   616
       - CT_ARCH_WITH_TUNE    : the gcc ./configure switch to select scheduling                 ( "--with-tune=${CT_ARCH_TUNE}"   )
yann@767
   617
       - CT_ARCH_WITH_FPU     : the gcc ./configure switch to select FPU type                   ( "--with-fpu=${CT_ARCH_FPU}"     )
yann@767
   618
       - CT_ARCH_WITH_FLOAT   : the gcc ./configure switch to select floating point arithmetics ( "--with-float=soft" or /empty/  )
yann@391
   619
   + provides:
yann@391
   620
     - optional
yann@767
   621
     - the environment variables to pass to the cross-gcc to build target binaries (defaults)
yann@391
   622
       - CT_ARCH_ARCH_CFLAG   : the gcc switch to select architecture level                     ( "-march=${CT_ARCH_ARCH}"            )
yann@456
   623
       - CT_ARCH_ABI_CFLAG    : the gcc switch to select ABI level                              ( "-mabi=${CT_ARCH_ABI}"              )
yann@391
   624
       - CT_ARCH_CPU_CFLAG    : the gcc switch to select CPU instruction set                    ( "-mcpu=${CT_ARCH_CPU}"              )
yann@391
   625
       - CT_ARCH_TUNE_CFLAG   : the gcc switch to select scheduling                             ( "-mtune=${CT_ARCH_TUNE}"            )
yann@391
   626
       - CT_ARCH_FPU_CFLAG    : the gcc switch to select FPU type                               ( "-mfpu=${CT_ARCH_FPU}"              )
yann@391
   627
       - CT_ARCH_FLOAT_CFLAG  : the gcc switch to choose floating point arithmetics             ( "-msoft-float" or /empty/           )
yann@391
   628
       - CT_ARCH_ENDIAN_CFLAG : the gcc switch to choose big or little endian                   ( "-mbig-endian" or "-mlittle-endian" )
yann@391
   629
     - default to:
yann@391
   630
       see above.
yann@767
   631
   + provides:
yann@767
   632
     - optional
yann@767
   633
     - the environement variables to configure the core and final compiler, specific to this architecture:
yann@767
   634
       - CT_ARCH_CC_CORE_EXTRA_CONFIG   : additional, architecture specific core gcc ./configure flags
yann@767
   635
       - CT_ARCH_CC_EXTRA_CONFIG        : additional, architecture specific final gcc ./configure flags
yann@767
   636
     - default to:
yann@767
   637
       - all empty
yann@767
   638
   + provides:
yann@767
   639
     - optional
yann@767
   640
     - the architecture-specific CFLAGS and LDFLAGS:
yann@767
   641
       - CT_ARCH_TARGET_CLFAGS
yann@767
   642
       - CT_ARCH_TARGET_LDFLAGS
yann@767
   643
     - default to:
yann@767
   644
       - all empty
yann@628
   645
yann@903
   646
You can have a look at "config/arch/arm.in" and "scripts/build/arch/arm.sh" for
yann@903
   647
a quite complete example of what an actual architecture description looks like.
yann@901
   648
yann@890
   649
Kernel specific |
yann@890
   650
----------------+
yann@890
   651
yann@890
   652
A kernel is defined by:
yann@890
   653
yann@890
   654
 - a human-readable name, in lower case letters, with numbers as appropriate.
yann@890
   655
   The underscore is allowed; space and special characters are not (although
yann@890
   656
   they are internally replaced with underscores.
yann@890
   657
     Eg.: linux, bare-metal
yann@890
   658
 - a file in "config/kernel/", named after the kernel name, and suffixed with
yann@890
   659
   ".in".
yann@890
   660
     Eg.: config/kernel/linux.in, config/kernel/bare-metal.in
yann@901
   661
 - a file in "scripts/build/kernel/", named after the kernel name, and suffixed
yann@901
   662
   with ".sh".
yann@901
   663
     Eg.: scripts/build/kernel/linux.sh, scripts/build/kernel/bare-metal.sh
yann@890
   664
yann@890
   665
The kernel's ".in" file must contain:
yann@890
   666
 > an optional lines containing exactly "# EXPERIMENTAL", starting on the
yann@890
   667
   first column, and without any following space or other character.
yann@890
   668
   If this line is present, then this kernel is considered EXPERIMENTAL,
yann@890
   669
   and correct dependency on EXPERIMENTAL will be set.
yann@901
   670
yann@890
   671
 > the config option "KERNEL_%kernel_name%" (where %kernel_name% is to be
yann@890
   672
   replaced with the actual kernel name, with all special characters and
yann@890
   673
   spaces replaced by underscores).
yann@890
   674
   That config option must have *neither* a type, *nor* a prompt! Also, it can
yann@890
   675
   *not* depends on EXPERIMENTAL.
yann@890
   676
     Eg.: KERNEL_linux, KERNEL_bare_metal
yann@890
   677
   + mandatory:
yann@890
   678
       defines a (terse) help entry for this kernel.
yann@890
   679
       Eg.:
yann@890
   680
         config KERNEL_bare_metal
yann@890
   681
           help
yann@890
   682
             Build a compiler for use without any kernel.
yann@890
   683
   + optional:
yann@890
   684
       selects adequate associated config options.
yann@890
   685
       Eg.:
yann@890
   686
         config KERNEL_bare_metal
yann@890
   687
           select BARE_METAL
yann@890
   688
           help
yann@890
   689
             Build a compiler for use without any kernel.
yann@890
   690
yann@890
   691
 > other kernel specific options, at your discretion. Note however that, to
yann@890
   692
   avoid name-clashing, such options should be prefixed with
yann@890
   693
   "KERNEL_%kernel_name%", where %kernel_name% is again tp be replaced with
yann@890
   694
   the actual kernel name.
yann@890
   695
   (Note: due to historical reasons, and lack of time to clean up the code,
yann@890
   696
    I may have left some config options that do not completely conform to
yann@890
   697
    this, as the kernel name was written all upper case. However, the prefix
yann@890
   698
    is unique among kernels, and does not cause harm).
yann@890
   699
yann@901
   700
The kernel's ".sh" file API:
yann@901
   701
 > is a bash script fragment
yann@901
   702
yann@965
   703
 > defines the function CT_DoKernelTupleValues
yann@965
   704
   + see the architecture's CT_DoArchTupleValues, except for:
yann@965
   705
   + set the environment variable CT_TARGET_KERNEL, the kernel part of the
yann@965
   706
     target tuple
yann@965
   707
   + return value: ignored
yann@965
   708
yann@901
   709
 > defines the function "do_print_filename":
yann@901
   710
   + parameters: none
yann@901
   711
   + environment:
yann@901
   712
      - all variables from the ".config" file,
yann@901
   713
   + return value: ignored
yann@901
   714
   + behavior: output the kernel's tarball filename, with adequate suffix,
yann@901
   715
     on stdout.
yann@901
   716
       Eg.: linux-2.6.26.5.tar.bz2
yann@901
   717
yann@901
   718
 > defines the function "do_kernel_get":
yann@901
   719
   + parameters: none
yann@901
   720
   + environment:
yann@901
   721
      - all variables from the ".config" file.
yann@901
   722
   + return value: 0 for success, !0 for failure.
yann@901
   723
   + behavior: download the kernel's sources, and store the tarball into
yann@901
   724
     "${CT_TARBALLS_DIR}". To this end, a functions is available, that
yann@901
   725
     abstracts downloading tarballs:
yann@901
   726
     - CT_DoGet <tarball_base_name> <URL1 [URL...]>
yann@901
   727
       Eg.: CT_DoGet linux-2.6.26.5 ftp://ftp.kernel.org/pub/linux/kernel/v2.6
yann@901
   728
     Note: retrieving sources from svn, cvs, git and the likes is not supported
yann@901
   729
     by CT_DoGet. You'll have to do this by hand, as it is done for eglibc in
yann@901
   730
     "scripts/build/libc/eglibc.sh"
yann@901
   731
yann@901
   732
 > defines the function "do_kernel_extract":
yann@901
   733
   + parameters: none
yann@901
   734
   + environment:
yann@901
   735
      - all variables from the ".config" file,
yann@901
   736
   + return value: 0 for success, !0 for failure.
yann@901
   737
   + behavior: extract the kernel's tarball into "${CT_SRC_DIR}", and apply
yann@901
   738
     required patches. To this end, a function is available, that abstracts
yann@901
   739
     extracting tarballs:
yann@901
   740
     - CT_ExtractAndPatch <tarball_base_name>
yann@901
   741
       Eg.: CT_ExtractAndPatch linux-2.6.26.5
yann@901
   742
yann@901
   743
 > defines the function "do_kernel_headers":
yann@901
   744
   + parameters: none
yann@901
   745
   + environment:
yann@901
   746
      - all variables from the ".config" file,
yann@901
   747
   + return value: 0 for success, !0 for failure.
yann@901
   748
   + behavior: install the kernel headers (if any) in "${CT_SYSROOT_DIR}/usr/include"
yann@901
   749
yann@901
   750
 > defines any kernel-specific helper functions
yann@901
   751
   These functions, if any, must be prefixed with "do_kernel_%CT_KERNEL%_",
yann@901
   752
   where '%CT_KERNEL%' is to be replaced with the actual kernel name, to avoid
yann@901
   753
   any name-clashing.
yann@901
   754
yann@901
   755
You can have a look at "config/kernel/linux.in" and "scripts/build/kernel/linux.sh"
yann@903
   756
as an example of what a complex kernel description looks like.
yann@901
   757
yann@620
   758
Adding a new version of a component |
yann@476
   759
------------------------------------+
yann@476
   760
yann@476
   761
When a new component, such as the Linux kernel, gcc or any other is released,
yann@476
   762
adding the new version to crosstool-NG is quite easy. There is a script that
yann@476
   763
will do all that for you:
yann@476
   764
  tools/addToolVersion.sh
yann@476
   765
yann@476
   766
Run it with no option to get some help.
yann@381
   767
yann@203
   768
Build scripts |
yann@476
   769
--------------+
yann@203
   770
yann@203
   771
To Be Written later...