Block unsupported processors according to gcc version.
Also remove the comments since we now hide them according to this.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Bump default snapshot gcc version to 4.8-20120429 so that it is newer
than our latest supported version (4.7.0 release).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Decimal floats were introduced circa gcc-4.2 or -4.3, and requires
the floating-point environement fenv.h in the C library.
The uClibc .config file used by crosstool-NG to build uClibc is the
same as used by the internal buildroot mechanism, and explcitly
disables fenv support.
The quick workaround is to simply disable decimal floats in all
crosstool-NG config files.
In the long run, it might be better to check this situation, and/or
add code and/or options in crosstool-NG to handle this (but it is
much more involved, and this workaround is sane).
Reported-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
SUPPORT_LIB_DIR would get resolved to the main buildroot directory for
external toolchains without C++ support, as:
- gcc -print-file-name=<nonexisting-file> returns <nonexisting-file>
- readlink -f <nonexisting-file> returns $PWD/<nonexisting-file>
So fix it by ensuring output of gcc -print-file-name actually exists
before using it.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Update the version of crosstool-Ng used, bump to 1.15.2.
Also, update the bundled config files to match the new version.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Commit 0729b544b3 (Improve external toolchain logic to support IA32
Sourcery CodeBench toolchain) and e1f0804cc (external-toolchain: add
support for recent Linaro toolchains) changed the interface of
copy_toolchain_lib_root, but ctng wasn't updated so libraries weren't
copied to the target.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
In order to use locale support on a Linux system, you need locale data
to be present:
* on a (e)glibc based system, this data is typically in the
/usr/lib/locale/locale-archive file, which can be created and
extended using the localedef program
* on an uClibc based system, the set of supported locales is defined
at build time by an uClibc configuration option.
This patch implements generating locale data for the following cases:
* Internal toolchain
* External toolchain based on (e)glibc. uClibc external toolchains
are not supported, because with uClibc, the set of supported
locales is defined at build time. CodeSourcery and Linaro
toolchains have been tested, Crosstool-NG toolchains are believed
to work properly as well.
* Toolchains built using the Crosstool-NG backend, but only (e)glibc
toolchains.
This feature was runtime tested with internal uClibc toolchain,
CodeSourcery ARM toolchain and Linaro ARM toolchain, thanks to a
simple C program that shows the data and a gettext translated message.
Note that this option differs from the "purge locales" option, which
is responsible for removing translation files and other locale stuff
installed by packages. At some point in the future, we may want to
clarify the respective roles of those options.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
When the crosstool-ng backend is used, host-gawk is built as a
dependency of the crosstool-ng package, and therefore an host 'gawk'
binary is installed in $(HOST_DIR).
When the target gdb package is also selected, this unfortunately leads
to a build failure, as reported on
http://buildroot.humanoidz.org/results/f19c0499d08212d8b5100fa9434e1197092957db/build-end.log.
The problem is that the ./configure of gdb detects gawk in the PATH,
but at compile time, it fails to find gawk. This is due to the fact
that the gdb compilation process is started without the correct path.
This patch fixes this by passing $(TARGET_MAKE_ENV) in the environment
of the gdb compilation process.
A better fix would be to switch gdb to the AUTOTARGETS infrastructure
in the future.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Starting from 2012.03, the Linaro toolchains have separated the GCC
support libraries (libstdc++, libgcc_s) from the sysroot itself. So we
no longer have the case where all libraries are inside the sysroot, as
we had for all the previously supported toolchains.
Therefore, we add some logic to detect if such a separate directory is
used for GCC support libraries, and if it's the case, we make sure
that we take into account this directory when creating our own
sysroot, and when copying libraries to the target filesystem.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Enable uClibc 0.9.33.1 for the SH architecture.
Builds and works fine with the qemu sample config.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Enable AI_ADDRCONFIG support in uClibc otherwise we don't get the
ifaddrs.h header installed and thus lack getifaddrs support.
Quite useful for samba for example so we can use interfaces=eth0 statements in
the configuration file rather than klunkier
interfaces=192.168.1.1/255.255.255.0 ones.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The support for mklibs has been marked broken since more than a year
and nobody cared to bring it up to a working state.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Re-integrate in gcc-uclibc-4.x.mk things from
toolchain/gcc/Makefile.in that were completely gcc-specific. There was
no reason to pull that when building with other backends than the
internal one.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The BR2_CONFIGURE_DEVEL_SYSROOT, BR2_CONFIGURE_STAGING_SYSROOT and
BR2_CONFIGURE_BUILD_TOOLS were used only in a few places, and it is in
fact clearer to just use their value in the various places they are
used.
The ultimate goal is to get rid of the toolchain/Makefile.in file.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Add gcc 4.7.0 to the toolchain options.
[Peter: drop 0001-toolchain-gcc-add-4.7.x-series.txt]
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Switch from the strict UCLIBC_HAS_CTYPE_ENFORCED=y to the less
restrictive UCLIBC_HAS_CTYPE_CHECKED=y since it breaks dialog under
certain circumstances (bug #5018).
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The big-endian microblaze external toolchain is missing vital
characteristics being set, such as LARGEFILE support, IPV6 and so on.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: Stephan Hoffmann <sho@reLinux.de>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The 3.3 kernel headers define the umode_t type within a __KERNEL__
preprocessor ifdef region. This results in a broken kernel header in the
buildroot toolchain.
[Peter: This is discussed upstream here:
https://bugzilla.kernel.org/show_bug.cgi?id=42986
Long term socat/syslinux should stop using this header, but this hasn't
been fixed upstream yet]
Signed-off-by: Simon Dawson <spdawson@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
It adds very little size overhead as the functions are just wrappers
around utmp, and E.G. systemd needs it.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
While there's some microblaze support in mainline gcc from 4.6.x,
there still seems to be something missing with the uClibc support, so
disable these for now.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Alvaro G. M <alvaro.gamez@hazent.com>
Tested-by: Stephan Hoffmann <sho@relinux.de>
Downloading Microblaze LE toolchain works on a clean install
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
This modifies the definition of DOWNLOAD to receive two arguments:
the first one is the full URL of the file to download, whereas the second
(and optional) is the name the file will have once downloaded.
Same thing with the SOURCE_CHECK_WGET and SCP functions.
All calls to these functions have been changed to the shortest form of
the new API, except for toolchains acquisition. Since there is quite a
number of different toolchains this call to DOWNLOAD is better set to the
generic one.
Signed-off-by: Alvaro G. M <alvaro.gamez@hazent.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Tested-by: Stephan Hoffmann <sho@relinux.de>
Downloading Microblaze LE toolchain works on a clean install
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
When binutils was converted to the package infrastructure (commit
009407e6b), the variable that elf2flt uses to find the binutils
libraries disappeared. So use HOST_BINUTILS_DIR instead of
BINUTILS_DIR1.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
When an external toolchain is used, it is very likely that it contains
a pre-built version of a gdbserver that has the same version as the
cross-gdb included in the external toolchain. So, we now provide an
option that allows to copy this pre-built gdbserver to the target.
As the location of the gdbserver in the external toolchain is not
standardized, we only support the CodeSourcery and Crosstool-NG
layouts for the moment. Other locations can be added later.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Without the -m64, choosing the x86_64 architecture with a Sourcery external
toolchain will result in a 32-bit rootfs.
Also simplified the help text to reflect this change.
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Linaro has just released a new pre-built toolchain, available as a
tarball, which is a pure toolchain (only the C library is
included). This makes this new Linaro 2012.01 toolchain usable in
Buildroot, so let's integrate the support for it.
In addition to simply adding the new external toolchain at the usual
locations, this patch allows need to adapt a few things to support
Linaro toolchains. Most toolchains store their libraries in the "lib/"
or "usr/lib" directories relative to the toolchain. Buildroot
toolchains on the other hand, store the libraries in the
"usr/<target-name>/lib" directory. And the Linaro toolchain has
choosen to use the "lib/<target-name>" directory. Therefore, this
patch adjust:
* The logic to search a particular library when that library needs to
be copied to the target directory
* The logic to deduce the sysroot directory from the libc.a file
location in the toolchain: removing "(usr/?)lib(64?)" is no longer
sufficient, we need to take into account the "lib/<target-name>/"
case.
Since the Linaro toolchain generates code for Cortex-A processors
only, the selection of this toolchain is limited to Cortex-A8 and
Cortex-A9.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The current check for uClibc toolchain was verifying that a
ld-uClibc.so dynamic loader was present. However, with static-only
uClibc toolchains, this does not work. Instead, we check for an
uClibc-specific header file in the sysroot.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
When the mechanism that allows Buildroot to download external
toolchains automatically was added, all the sanity checks on the
external toolchains were not performed. This commit re-enables those
checks that we already do on external toolchains that are not
downloaded/extracted by Buildroot. This makes the toolchain checks
more consistent accross various configurations.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Each multilib variant need to be selected using a special combination
of flags, requiring specific choices of the Buildroot options. This
commit documents those configuration choices to make it easier to use
the various multilib variants.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
We are going to add one more ARM Sourcery toolchain version, so it's
time to remove the oldest version.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The BR2_TARGET_OPTIMIZATION flags were not used by the external
toolchain wrapper, which broke the multilib selection logic of
multilib external toolchains. It also simplifies the compilation of
external programs since all flags are properly passed automatically by
the toolchain wrapper.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
When an external toolchain has multiple variants organized in
sub-directories, Buildroot only copies the selected sysroot and not
all sysroots. In order to make this work, Buildroot creates a symbolic
link of the name of the original selected sysroot to the main sysroot
to trick the compiler so that it finds its libraries at the expected
location.
I.e, if the toolchain as the following organization (example take on
the ARM CodeSourcery toolchain) :
. for ARMv5T
armv4 for ARMv4T
thumb2 for ARMv7-A/Thumb
and ARMv4T is selected, then Buildroot will copy the contents of
armv4t/ from the toolchain into its $(STAGING_DIR) and then create a
$(STAGING_DIR)/armv4t symbolic link to $(STAGING_DIR).
However, our logic to do so only works when there was one directory
level for multilib sysroots. But in the MIPS CodeSourcery toolchain
there are multiple levels. For example, the MIPS16 soft-float
little-endian sysroot variant is in mips16/soft-float/el/ compared to
the main sysroot.
This patch improves our logic to support this case. The logic is a bit
more complicated as we don't want to create a symbolic link to an
absolute path, but a symbolic link to a relative path, because we want
the host/ directory to be relocatable.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The IA32 Sourcery CodeBench toolchain has a relatively special
structure, with the following multilib variants:
* Intel Pentium 4, 32 bits, the multilib variant is in ./ relative to
the main sysroot, with the libraries in the lib/ directory.
* Intel Xeon Nocona, 64 bits, the multilib variant is in ./ relative
to the main sysroot, with the libraries in the lib64/ directory.
* Intel Atom 32 bits, the multilib variant is in atom/ relative to
the main sysroot, with the libraries in the lib/ directory.
* Intel Core 2 64 bits, the multilib variant is in core2/ relative to
the main sysroot, with the libraries in lib64/ directory.
So the first two variants are in the same sysroot, only the name of
the directory for the libraries is different.
Therefore, we introduce a new ARCH_LIB_DIR variable, which contains
either 'lib' or 'lib64'. This variable is defined according to the
location of the libc.a file for the selected multilib variant, and is
then used when copying the libraries to the target and to the staging
directory.
In addition to this, we no longer use the -print-multi-directory to
get the ARCH_SUBDIR, since in the case of the 64 bits variants of this
toolchain, it returns just '64' and not a real path. Instead, we
simply compute the difference between the arch-specific sysroot and
the main sysroot.
We also take that opportunity to expand the documentation on the
meaning of the different variables.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
As suggested by Arnout Vandecappelle, move toolchain/dependencies to
support/dependencies, as it really is not toolchain-specific anymore.
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
With default values so it doesn't stop build and ask user. Also disable
2.4 modules support by default like upstream does.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Similar to how we do for target (ee39d53ce3ee (Fix GDB BFD test linking)).
Gdb comes with an embedded copy of libiberty, but binutils also installs
libiberty.a into HOST_DIR. The gdb configure script tries to link against
this one rather than the gdb version when it checks for ELF support.
This may fail if those versions are not compatible, leading to obscure
error messages from gdb at runtime such as:
I'm sorry, Dave, I can't do that. Symbol format `elf32-$ARCH' unknown.
Fix it by forcing ELF support.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Bump 3.0.x series to 3.0.17, 3.1.x series to 3.1.9 and 3.2.x series to
3.2.1
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Has been marked as broken since July 2010 (39e6ba1b), and nobody has
stepped up to support it, so now finally remove it completely.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Add the ability for buildroot to build an SPE ABI enabled toolchain.
This is mandatory for e500v1/v2 cores since they don't support classic
FPU mode as the e500mc does.
Useful for Freescale's PowerQUICC III and single/dual-core QorIQ
line of processors.
The new TARGET_ABI variable is used rather than TARGET_CFLAGS for
uclibc's UCLIBC_EXTRA_CFLAGS to avoid breakish CFLAGS leaking in, a
good example being -mthumb for ARM.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
