1. On SuperH, configuring GCC with explicit variant of the CPU
(like "sh4") limits the default set of multilibs to just that CPU
and requires --with-multilib-list to change. Allow for "unspecified"
variant, so that we can defer to GCC to determine the list.
2. Support toolchains with both endiannesses at the same time.
3. Add a SuperH/newlib sample
4. Add more flags processing for uClibc
Signed-off-by: Alexey Neyman <stilor@att.net>
Also:
- Move companion_* to comp_* to match the kconfig symbols
- Replace bootstrap with former gen-versions.sh
- Fold *.in.2 into their respective first parts; this moves common
options to the end - if it is undesirable, inclusion of *.in
can be moved where *.in.2 used to be (but that will also move
version selection after common options).
- Retire addToolVersion.sh (may later replace with a more
comprehensive script that tries to download the added tarballs,
copy the patches and try to apply them, and create a version.desc).
Signed-off-by: Alexey Neyman <stilor@att.net>
Instead, prepare the files as a part of bootstrap and install them.
This avoids rebuilding these files in each working directory; they
don't change anyway as they are generated from the same installed
source.
Signed-off-by: Alexey Neyman <stilor@att.net>
It is only used if this libc flavor uses a multilib iterator (and not
determines the multilibs itself). This class currently includes glibc,
uClibc, musl - but they explicitly select CC_CORE_PASSES_NEEDED anyway.
Signed-off-by: Alexey Neyman <stilor@att.net>
It turns out buildroot does not currently accept a toolchain where a dynamic
linker does not reside in the multi-os-directory. Unfortunately this is
how glibc installs itself on AArch64 without any extra tricks.
So, provide an option to force everything into /lib or /usr/lib; patch to
buildroot will be worked on separately.
Signed-off-by: Alexey Neyman <stilor@att.net>
... these are apparently not needed with the current kconfig and only
result in warnings like "SYMBOL changed state" and "reassigning SYMBOL".
Perhaps, it was necessary to run kconfig without first generating
config.gen? But now all the targets that invoke $(CONF) have
`config_files` as a dependency.
Signed-off-by: Alexey Neyman <stilor@att.net>
On some arches (e.g. MIPS) the options like -mabi do not work if
specified more than once (see the comment in 100-gcc.sh). Therefore,
we need to determine which of the options produced by <arch>.sh can
be passed to multilib builds and which must be removed (i.e., which
options vary among the multilibs).
This presents a chicken-and-egg problem. GCC developers, in their
infinite wisdom, do not allow arbitrary multilib specification to be
supplied to GCC's configure. Instead, the target (and sometimes some
extra options) determine the set of multilibs - which may include
different CPUs, different ABIs, different endianness, different FPUs,
different floating-point ABIs, ... That is, we don't know which parts
vary until we build GCC and ask it.
So, the solution implemented here is:
- For multilib builds, start with empty CT_ARCH_TARGET_CFLAGS/LDFLAGS.
- For multilib builds, require core pass 1. Pass 1 does not build any
target binaries, so at that point, our target options have not been
used yet.
- Provide an API to modify the environment variables for the steps that
follow the current one.
- As a part of multilib-related housekeeping, determine the variable
part of multilibs and filter out these options; pass the rest into
CT_TARGET_CFLAGS/LDFLAGS.
This still does not handle extra dependencies between GCC options (like
-ma implying -mcpu=X -mtune=Y, etc.) but I feel that would complicate
matters too much. Let's leave this until there's a compelling case for
it.
Also, query GCC's sysroot suffix for targets that use it (SuperH,
for example) - the default multilib may not work if the command line
specifies the default option explicitly (%sysroot_suffix_spec is not
aware of multilib defaults).
Signed-off-by: Alexey Neyman <stilor@att.net>
The configure error mentioned in the commits that disable
--with-arch/--with-tune when --with-cpu is specified is specific to ARM
builds. For other architectures, it makes sense (hey, it even makes
sense for ARM - read 'info gcc'! - it is just a shortcoming of ARM
build, apparently).
Thus, add an arch setting ARCH_EXCLUSIVE_WITH_CPU (currently, set only
for ARM) that will trigger current behavior. Permit combinations of
--with-arch/cpu/tune for other arches.
Signed-off-by: Alexey Neyman <stilor@att.net>
This commit adds support for the Atmel AVR 8-bit RISC architecture.
This is the first 8-bit architecture to be added to crosstool-ng so the
configuration options for 8-bit architectures are added here as well.
gcc has had support for AVR for quite a while, at least since the 4.3
series for the currently popular ATmega microcontroler series.
The AVR architecture only supports bare-metal toolchains.
gcc for the AVR 8-bit architecture, usually referred to as avr-gcc, is
commonly used in conjunction with the avr-libc library which provides
additional resources for the Atmel AVR 8-bit microcontrollers.
avr-gcc can also be found as a supported package in some recent Linux
distributions.
This commit also closes#66
Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Similarly to what we've just done to prevent both --with-arch and
--with-cpu, we do the same to prevent using both --with-cpu and
--with-tune at the same time, since --with-cpu should fully imply
the CPU to tune for (and gcc now errors out when both are specified.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Bryan Hundven <bryanhundven@gmail.com>
Normally, a specific CPU fully implies the architecture level. For
example, a cortec-a8 is forcibly an armv7, so spwecifying both is
redundant, and even dangerous (as incompatible values may be passed).
So far, gcc was pretty happy when both were specified at the same time,
and some time ago, it started being a warning, and only recently was it
turned into a hard error.
So, hide the architecture level prompt when a CPU has been specified.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Add an ARCH_FLOAT_AUTO option to avoid passing float options
to gcc and allow it to choose
Signed-off-by: Cody P Schafer <dev@codyps.com>
[yann.morin.1998@free.fr: slightly reword the commit log]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Message-Id: <088489641f4790262685.1399687813@localhost>
Patchwork-Id: 347584
For some architectures, it is legit to have an alternate value in the
'architecture' part of the tuple. For example:
armv5te-*
armv7a8-*
Besides, some packages expect the tuple to reflect the arch variant
(eg. openMPI) to detect the variant's capabilities (eg. atomic
primitives).
This patch adds an option for the user to specify a suffix to be added
to the arch-part of the tuple.
Signed-off-by: Willy Tarreau <w@1wt.eu>
Message-ID: <20130120225822.GS6838@1wt.eu>
Patch-Id: 213994
[yann.morin.1998@free.fr: make it a suffix, not an override]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
It's been some time now we've had those features, so unmark them
being experimental.
It does not mean everything is perfect, but may gather some more
testing of those features.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Re-order the three entries in the float choice, in a more sensible
order, ie:
- all hard-float options come first, then soft-float
- options that use the FPU are marked so: hard and softfp
- options that do not use the FPU are marked so: software
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Add a toolchain-wide option to enable multilib.
This is currently a noop, and will be implemented
in subsequent patches for each impacted components.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Do not prompt for the type of floating-point support, if the
architecture did not explicitly stated that it did support it.
Reported-by: Morten Thunberg Svendsen <mts@doredevelopment.dk>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Changeset #149c33923f47 broke the architectures that do not
support the --with-float=X ./configure flag (in gcc). For example,
x86_64 does not support it.
Add a new blind config option that architectures can set to tell
they support floating point selection.
Reported-by: Morten Thunberg Svendsen <mts@doredevelopment.dk>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
First, 'SUPPORT' should be spelled 'SUPPORTS'.
Second, 'SUPPORT_XXX' really means 'supports --with-xxx', so rename the
affected options accordingly. Update the affected archs to match the new
namings.
Reported-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Some architectures support a mixed hard/soft floating point, where
the compiler emits hardware floating point instructions, but passes
the operands in core (aka integer) registers.
For example, ARM supports this mode (to come in the next changeset).
Add support for softfp cross compilers to the GCC and GLIBC
configuration. Needed for Ubuntu and other distros that are softfp.
Signed-off-by: Michael Hope <michael.hope@linaro.org>
[yann.morin.1998@anciens.enib.fr: split the original patch]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
With the upcoming softfp support, the case..esac test would become
a bit convoluted if it were to test three different booleans.
Introduce a new blind string config option that defaults to the
selected floating point type used.
Signed-off-by: Michael Hope <michael.hope@linaro.org>
[yann.morin.1998@anciens.enib.fr: split the original patch]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
kconfig bools are disabled by default, so specifying 'default n' is useless and
noisy. This patch removes all occurrences of 'default n'.
Signed-off-by: "Benoît THÉBAUDEAU" <benoit.thebaudeau@advansee.com>
Re-organise the sub-menu so that:
- the archs list comes first,
- followed by archs generic options
- followed by archs specific options
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Move the arch-specific options to the second-part of
the generated files, so they appear after the generic
options, but before the optimisations.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Some components have configuration options that can depend on
generic options, so they should go below those.
uClibc for example:
- has its own options (wchar...)
- uses the generic options (threads...)
- if linuxthreads chosen, offers two impls
So we need to be able to split the components options in 2,
one part that is above the generic options, and one part that
ends up below the generic options.
