gcc: add patch to fix EABI for armv4t

As pointed out by Martin GUY, gcc incorrectly generates armv5t
instrcutions for EABI, even for cores that are an armv4t.
The new patch (for the 4.3 series) fixes the problem by downgrading
the default CPU for EABI to being an armv4t core.

kconfig: allow stdin/stdout redirection

Allow stdin/stdout redirection for the CLI conf (not mconf).
This allows to recall a sample and automatically apply the defaults
to new configuration option, with something like the following:
yes "" |ct-ng "sample_name"

docs: add a tutorial on how to build a toolchain on Mac OS-X

Add a step-bystep tutorial to build a cross-toolchain on Mac OS-X.

companion libs: add latest CLooG/PPL versions

Add the latest 0.15.{4,5,6,7} CLoog/PPL.

samples: show the PPL, GLoog/ppl and MPC versions

When displaying a sample, do show the PPL, CLoog/ppl and MPC versions

samples: fix displaying selected tools

Do display selected tools / debug facilities when displaying a sample.

samples: replace usage of echo with printf

Using printf instead of echo allows easier and better formatting.

tools wrapper: fix building

Remove the build tools only after the wrapper is built.
Use the corect C compiler to build the tools wrapper.
Use the correct log level.

crosstool.sh.in: better mesage when creating script-overrides

log functions: fix CT_DoLog and CT_DoExecLog

$@ and $* are different when in double quotes; use $* to print the message.

tools wrapper: move choice selection to a more appropriate place

Move the tools wrapper choice selection down to the companion libraries
sub-menu, to avoid the user going back and forth in the menu.

tools wrapper: fix config dependency

The tools wrapper is not needed only for canadian crosses, but also
for every other type of toolchain.

docs: typo in overview.txt

TODO: update

Add TODO list for m4, autoconf, automake and libtool.
Building our own versions would remove burden from the users
who have older versions on their distributions, and are not
ready/able/allowed to upgrade.

config.{gues,sub}: update

config: move the patch choice to the extract section

Choosing the origin of the patches to apply is best
done in the 'extract' section.

config: add fallback to patch order

Add the possibility to fallback to either bundled or local patches
if local or bundled are missing.

config: add the local,bundled patch order

This ordering first applies the local patches, then the bundeld ones.

config: make selecting the patch origin a choice rather than a bool

functions: do not over-detect error when patching

Merge.

tools wrapper: fix error due to unused argument

In the C wrapper, the argc argument is not used, causing an error
(as we treat warnings as errors). Use a dummy allocation to get
rid of the warning.

tools wrapper: correctly install the C wrapper

When building the C wrapper, specify the source file to be compiled.

config: move the "build shared libraries" option to the OS menu

The "Build shared libraries" config option is dependant on the type of
"Target OS".

Moving this options to the "Target OS" sub-menu is also better in the user
perspective: he/she no longer needs to go back and forth to see if he/she
missed any option.

config: make "Build shared libraries" depend on KERNEL_SUPPORTS_SHARED_LIBS

Now that all kernels that support shared libraries select the correct config
option, we can change the dependency rules of SHARED_LIBS accordingly.

config: Linux kernel selectes shared libraries support

The Linux kernel supports using shared libraries, so prompt the user.

config: introduce the KERNEL_SUPPORTS_SHARED_LIBS silent config option

The target OSes will be able to select this option if they support shared
libraries, instead of having "Build shared libaries" depend on each OS.

config: swap the debug and tools sub-menus

ltrace, in the debug sub-menu, selects libelf, in the tools sub-menu.
Inverse the order of the two sub-menus, so that the user does not have
to go back and forth between the two sub-menus.

config: move down companion libraries sub-menu

Move the companion libraries sub-menu down the main menu.
That way, the user does not need to go back and forth in the menu
to change options set by the different components that select the
companion libraries (binutils, gcc, gdb).

companion-libs: PPL/CLooG/MPC require GMP and MPFR

The PPL, CLoog/ppl and MPC libraries require GMP and MPFR.
Select GMP/MPFR appropriately.

tools wrapper: introduce the silent WRAPPER_NEEDED config option

Add the WRAPPER_NEEDED silent config option, that can be selected by
components that require it (companion libs so far).
Rely on this config option when deciding to install the wrapper,
instead of checking GMP/MPFR or PPL/CLoog/MPC.

Merge the C wrapper.

config: choose whether to use the shell or the C wrapper

Offer a config choice on whether to isntall the script wrapper, or
the compiled C wrapper. Update docs/overview.txt accordingly.

tool wrapper: add initial wrapper coded in C

Add an initial wrapper:
- find the realpath of the tool being called
- add the '.' in front of the tool name
- add the '/lib' dir to the base dir of the tool
- set and export LD_LIBRARY_PATH
- execve the real tool

overview: fix table of content, fix the hacker's way

table of content: add missing entry
the hacker's way: no need to "make install"

README: fix some space alignment

duma: fix downloading by forcing extension to .tar.gz

Downoading a non-existing file from sourceforge gives you a "200 OK"
and an index.html. As we try to retrieve a .tar.bz2 first, and duma
is bundled in a .tar.gz, we won't get appropriate content, so
just force the extension to avoid the problem.

Thanks to Ingmar Schraub <is@eseco.de> for pointing out the issue.

x86_64: is a MMU-aware arch, so select ARCH_USE_MMU

duma: fix version string.

DUMA uses '_' not '.' for major/minor/sub separation.

glibc: fix build error caused by incorrect variable assignment

During the conversion to using bash arrays, the glibc build script
was improperly converted, and contains an incorrect variable
assignment to the config_options array.

Merge the bash_array branch.

For every components where it makes sense, use bash arrays (instead
of a string with space-separated values) to store the options pased
to ./configure.

Allow setting --enable-cxx-flags on gcc ./configure

Some setups require one to pass extra CXX flags at the time of ./configure.
Make it easy.

Use bash arrays to store binutils config options

Make glibc's extra_config an array containing ./configure options

Change extra_config from a string to a array of options.

Make gdb's extra_config an array containing ./configure options

Change extra_config from a string to a array of options.

Make dmalloc's extra_config an array containing ./configure options

Change extra_config from a string to a array of options.

Make gcc's extra_config an array containing ./configure options

Change extra_config from a string to a array of options.

Make eglibc's extra_config an array containing ./configure options

Change extra_config from a string to a array of options.

configure: check for bash-3.1 or above

This is required to use the array variables that came with bash-3.1.

[libc-eglibc] Rewrite part of the code

Rewrite part of the code to better match the rest.
Most notably, rewrite handling of:
if [ ... ] && [ ... ]
to:
if [ ... -a ... ]

This has the positive side effect of calling "[" only once, although
"[" is probably a shell built-in.

[libc-eglibc] Fix testing for existing files

To test for existing files, use "[ -f blabla ]", not "[ -a blabla ]"
Checking for a file exsitence with "-a" is a bashism.
Althoug we _are_ using bash, it's disturbing as it can be misread as
the 'and' operator. Fix by using "-f".

samples: add eglibc-based toolchain for MIPS.

config: add Linux kernel 2.6.30.5

config: update Linux kernel long-term stable to 2.6.27.31.

Merge the AVR32 support branch to the default branch.

Removed debug line

Fix libc config

Misc AVR32 cleanups.

Changed copyright

Update README on how to send proper patches

Add a few lines of explanations on how to correctly setup
and use Mercurial to send proper patches.

Add sections to the README

Fix typo in README

Update the README with the new URL to the project homepage

The URL for the crosstool-NG homepage has changed. Update the
README accordingly. The old URL is still valid, though.

config/gcc: re-order versions

Move the gcc-4.3.4 version to its place.

gcc: add support for 4.3.4

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

gcc: add patch to fix PR37436

When compiling some C++ code, GCC 4.3.x fails with an internal
compiler error. The bug report is available at
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37436. The included patch
is the one that has been merged in the trunk of gcc.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>

config: make CONFIG_SHELL default to bash

Some components (eg. GMP) will fail to correctly build if
the CONFIG_SHELL is not bash (eg. ash or dash). So make bash
the default CONFIG_SHELL.

Keep ash as a possible selection, as future versions of those
currently /broken/ tools may come fixed wrt to CONFIG_SHELL
being POSIX-ly compliant.

Tagging release 1.4.2

1.4: update version to 1.4.2

[complib:mpfr] Fix building MPFR in some weird cases

The tmul test uses a compiled-in input file in $(srcdir).
The problem is that the Makefile passes it unquoted. The C code
tries to stringify it using clever macros, which may *usually* work.

In my case the source directory was named:
.../toolchain-powerpc-e500v2-linux-gnuspe-1.0-2.fc10/.../tests
And guess what? During testing I found out the program fails because
it tries to open:
.../toolchain-powerpc-e500v2-1-gnuspe-1.0-2.fc10/.../tests

Yes, CPP tokenized the macro before stringifying it and not surprisingly
the 'linux' part was converted to 1.
[on Fedora-10: cpp (GCC) 4.3.2 20081105 (Red Hat 4.3.2-7)]

So the attached patch simplify the macros and pass the path as string
from the Makefile.

Manually backported from 1449:8ad2773e7ae3

[configure] Fix automake version check

The configure script fails on automake-1.11 (in Fedora-11) since
it looks for 3-digit version number.

"Yann E. MORIN", added the following comment:

The check for the automake version is not against a 3-digit number,
but really against a 3-part version number, a-la 'x.y.z'. Versions
such as 1.10 and 1.11 are also valid.

Fix creating the .../lib/ directory structure in the non-sysroot case

The symbol link that is created in the sysroot directory only needs
to be made when the cross compiler is build with the sysroot option

Signed-off-by: Bart van der Meulen <bartvdrmeulen@gmail.com>
(transplanted from 0e49e6eddac4f5bbde6394eb3e6e36be75fc550c)

Also set CT_DEBUGROOT_DIR for non sysroot-ed toolchains.

In the non-sysroot-ed case, the debuf-root directory would not be set;
debug tools would have been installed God-only-knows-where...

Spotted by Bart van der Meulen <bartvdrmeulen@gmail.com>.
(transplanted from 634ffd14d936154464fd714ed29ed44c8e16ae02)

Add .hgignore file.

List all generated or temp files in .hgignore, so they
don't clutter the screen when running hg status.
(transplanted from befdb88324e18e1e80260906d0459a56ea0eb37a)

populate: fix installing dynamic linker 'ld.so'

The dynamic linker, ld.so, needs the execute bit to be set.
Detect tht the library being installed is in fact ld.so and
install it with 0755 instead of 0644.

Fix detecting src == dst.

Use a simpler command to copy src -> dst.

Also change echo to printf, get rid of 'echo -n', which is
highly non-portable.


-------- diffstat follows --------
/trunk/scripts/populate.in | 76 43 33 0 +++++++++++++++++++++++++++++-----------------------
1 file changed, 43 insertions(+), 33 deletions(-)
(transplanted from d7ddcb75e0f703e2ba6d17169167356389224870)

[configure] Use hg, instead of svn, to compute the version string

Now we moved from Subversion to Mercurial, use the hg command to
compute the version string.

Manually backported from 1409:7264ce426be4 plus 1430:933eee133cbf
as too much changes have been made to ./configure in the meantime.

[complib:mpfr] Fix building MPFR in some weird cases

The tmul test uses a compiled-in input file in $(srcdir).
The problem is that the Makefile passes it unquoted. The C code
tries to stringify it using clever macros, which may *usually* work.

In my case the source directory was named:
.../toolchain-powerpc-e500v2-linux-gnuspe-1.0-2.fc10/.../tests
And guess what? During testing I found out the program fails because
it tries to open:
.../toolchain-powerpc-e500v2-1-gnuspe-1.0-2.fc10/.../tests

Yes, CPP tokenized the macro before stringifying it and not surprisingly
the 'linux' part was converted to 1.
[on Fedora-10: cpp (GCC) 4.3.2 20081105 (Red Hat 4.3.2-7)]

So the attached patch simplify the macros and pass the path as string
from the Makefile.

[config] Merge CONFIG_SHELL selection.

[config] Warn againt using ash as CONFIG_SHELL

It apears that more and more ./configure scripts and Makefiles
make use of non-POSIXly correct shell constructs, that don't
work with ash.

For now, just warn the user against using ash, but keep it as
an option in case newer versions of the /broken/ components
come fixed, and we can again use ash, as it is in some cases
really faster than bash.

[config] Allow user to specify custom CONFIG_SHELL

In case the shell the user wants to use as CONFIG_SHELL is located in a
weird place (eg. /opt/bash/bin/bash), or is weirdly named (eg. bash-4),
let the user enter the patch to the shell.

[config] Add bash as a possible CONFIG_SHELL

On some systems and/or for some components, it may be necessary to
explicitly use bash as the shell used by ./configure and Makefiles.

[config] Change the way to select the CONFIG_SHELL

Transform the single boolean option to a 'choice' selection
with two options: "sh" and "ash".

This is needed to later add "bash" as override.

Merge the queue with all new component versions.

[cc-gcc] Add latest versions

Add gcc-4.4.1

[scripts-add-version] Handle gcc 4.4+

Handle gcc-4.4+ when adding a new gcc version.

[comp-lib-gmp] Add latest versions

Add 4.3.0 and 4.3.1.

[tools-libelf] Add latest version.

Add libelf 0.8.11, propagate patchset from 0.8.10.

[libc-eglibc] Add latest version

Add latest branch 2_10.

[debug-ltrace] Add latest version.

Add ltrace 0.5.2, propagate patchset from 0.5.1 and from Debian.

[debug-duma] Add latest versions

Add 2.5.15, propagate patchset from 2_5_14 up to 2_5_15.

[kernel-linux] Add latest versions

Add 2.6.29.{5,6} and 2.6.30{,.1}
Update 2.6.27.24 to 2.6.27.26 as long-term stable.

fix wrapper script for symlinks

The wrapper script placed around the target binaries when
using the companion libraries does not work for symbolic links
The wrapper scripts needs to follow the links before calling the
actual binary

Signed-off-by: Bart vdr. Meulen <bartvdrmeulen@gmail.com>

---

Add g++ to required tools

Because gmp is now always build with --enable-cxx,
g++ has become a required tool

Signed-off-by: Bart vdr Meulen <bartvdrmeulen@gmail.com>

---

[configure] Fix automake version check.

The configure script fails on automake-1.11 (in Fedora-11) since
it looks for 3-digit version number.

Patch fixed by "Yann E. MORIN", with the added comment:

The check for the automake version is not against a 3-digit number,
but really against a 3-part version number, a-la 'x.y.z'. Versions
such as 1.10 and 1.11 are also valid.

Require GNU awk at ./configure time.

Some components (specifically {,e}glibc) will not build when awk
is not GNU awk. Make ./configure check for GNU awk.

Make it build again for non-local use.

Using colons (:) in the version string creates Makefile problems:

Makefile:195: *** multiple target patterns. Stop.
make: *** [build] Error 2

Merge the canadian branch to default.

Now the preliminray support for canadian-crosses is done,
merge to default. The few remaining quirks, if any, will
be resolved on mainstream.

[samples] Add a linux -> mingw32 -> i686-linux canadian sample

Add a sample that builds on Linux a mingw32-hosted toolchain that
generates code for i686-none-linux-gnu.

Signed-off-by: Bart van der Meulen <bartvdrmeulen@gmail.com>

[samples] Don't mix canadian samples with non-canadian ones.

- save the canadian smples in their own way, so as not to
mix non-canadian samples with canadian ones
- list canadian samples with the host information

Cleanup the handling of canadian settings

Bring handling the 'canadian' settings as par with the existing
'cross' settings.

Add support code for the canadian-cross case.

Add implementation for a candadian build option already
present in crosstool in order to build a cross-compiler
where build != host != target

Signed-off-by: Bart van der Meulen <bartvdrmeulen@gmail.com>

[samples] Make saveSample.sh a little bit more generic

Prepare saving canadian samples by making the saveSample.sh script
a little bit more generic, using conditional code-paths (even if
there's currently a single code-path).

Don't use prefix/bin to install the buildtools.

Collect the build tools in a seperate folder in order to prevent accidental
calling our newly build tools.

Signed-off-by: Bart van der Meulen <bartvdrmeulen@gmail.com>

Fix MPFR build to recognise the Mingw32 case.

Once we have canadian in place, Mingw32 can be a legitimate host,
so we have to recognise that along with Cygwin.

Also fix recognising Cygwin hosts.

Signed-off-by: Bart van der Meulen <bartvdrmeulen@gmail.com>

Use configured shell in all build-tools wrappers

Abide by the user's wish to use dash as shell in
build-tools wrappers, instead of plain /bin/sh.

Fix creating the .../lib/ directory structure in the non-sysroot case

The symbol link that is created in the sysroot directory only needs
to be made when the cross compiler is build with the sysroot option

Signed-off-by: Bart van der Meulen <bartvdrmeulen@gmail.com>

Also set CT_DEBUGROOT_DIR for non sysroot-ed toolchains.

In the non-sysroot-ed case, the debuf-root directory would not be set;
debug tools would have been installed God-only-knows-where...

Spotted by Bart van der Meulen <bartvdrmeulen@gmail.com>.

[tools] Update config.guess and config.sub

Update from upstream.

Remove left-over debug lines.

Remove two lines of code that were commented out and are definitely not used.

Check paths sanity before they get used.

Don't use the PATHs before they get checked for sanity.

Better check for spaces in paths

Spaces are evil in paths. Print the path that contains a sapce.
Don't print computed paths, as they'rebased on provided paths,
and don't get space added into them.

[functions] Allow newlines (\n) in log messages

Allow caller to log messages with embedded newlines when calling
CT_DoLog and other logging functions (eg. CT_Abort...).

[finish-step] Fix the tools wrappers

Use the correct variable when iterating upon the tools
name to correctly link the tools to the wrapper.

Add .hgignore file.

List all generated or temp files in .hgignore, so they
don't clutter the screen when running hg status.

Merge the gcc-4.4 support.

[gcc] Merge the gcc-4.4 suport to the default branch, from 1407:83083566fdd0

[configure] Use hg to compute the version string

Now that we have migrated to hg from svn, it does not make sense
to keep using svn tools to compute the version string, so switch
to using mercurial tools (that are much more flexible, BTW).

[finish-step] Use local variable in finish-step

When wrapping the tools, use a local variable when iterating.

[companion-libs] Fix spaces in help entries

Some help entries were missing proper alignment-spaces.

[populate] enhance documentation

This patch adds explanations on how
populate works internally.