The big.LITTLE configurations can be optimised for by gcc, and a few
users wonder what they should choose when they have such CPUs.
Add new entries for those big.LITTLE configurations.
Note: the various combos were added in various gcc versions, but only
really worked in later versions:
Variant | Introduced in | First built in
----------+---------------+----------------
a15-a7 | 4.9 | 4.9
a17-a7 | 5 | 5
a57-a53 | 4.9 | 6
a72-a53 | 5 | 6
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Thomas De Schampheleire <patrickdepinguin@gmail.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
We have to specify the -mcpu value, even in 64-bit mode.
For AArch64, +fp and +simd are the default, so they are totally useless.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Whitespaces were searched using the following regex:
[ ]{1,}\t
and then manually removed in most of the cases. For
xserver_xorg-server.mk, tabs before backslashes were removed.
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This config option corresponds to the string returned by readelf for
the "Machine" field of the ELF header. It will be used to check if the
architecture of binaries built by Buildroot match the target
architecture.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The ARMv8 cores all support thumb2 instructions when running in aarch32 mode.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
A number of packages use BR2_ARM_CPU_HAS_NEON to know if the target handles
aarch32 neon instructions, which is only true for ARMv8 cores when they are
running in 32bit mode.
Notice: These cores do support neon-like instructions using a different
encoding in 64bit mode (it is a required part of ARMv8, similar to the FPU).
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Fixes:
http://autobuild.buildroot.net/results/5e6/5e67cc067a06f7364cde1a8393ea72608fe7fef1/
A number of packages use BR2_ARM_CPU_HAS_ARM to know if the target handles
classic A32 instructions, which is only true for ARMv8 cores when they are
running in 32bit mode.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Add two popular ARM64 cores to the list of supported cores: Cortex-A57
and Cortex-A72.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Adds the Cortex-A53 CPU to the target architecture variant choice. This
sets the toolchain to use Cortex-A53 as the target. The effect is that
various Cortex-A53 tunings are enabled for the compilation of packages.
Signed-off-by: Matt Flax <flatmax@flatmax.org>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
There's currently only one widely supported ABI for ARM64, called lp64,
so we define BR2_GCC_TARGET_ABI to the appropriate value.
Note that there is another ABI for ARM64 being worked on, ilp32, but its
support is not fully upstream in the kernel, so we're not adding support
for it for the moment.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The ARMv8 cores have a mandatory FPU unit called FP-ARMv8, so we:
- add a new hidden Config.in option for the availability of this
unit (BR2_ARM_CPU_HAS_FP_ARMV8)
- allow the selection of a possible choice in the "Floating point
strategy", and add two new choices: BR2_ARM_FPU_FP_ARMV8 and
BR2_ARM_FPU_NEON_FP_ARMV8.
- specify the -mfpu values for BR2_ARM_FPU_FP_ARMV8 and
BR2_ARM_FPU_NEON_FP_ARMV8 cases, when used on ARM 32 bits (-mfpu
doesn't exist on ARM64, instead -mcpu modifiers are used, so they
will be added on a per-core basis).
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
[yann.morin.1998@free.fr: drop the FP strategy dependency]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The floating point strategy currently depends on EABI || EABIHF. The
reason was that, wayback when we also supported OABI, we only exposed FP
for EABI or EABIHF, and hide it for OABI, which did not support FP.
It's been a while now that we do not support OABI, but the dependency
stuck all along.
Remove it as it is no longer needed, and is always true.
However, the choice is empty for AArch64, as we still have no entry for
their floating point strategy yet.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
In order to prepare the addition of ARM64 cores, add the blind
BR2_ARM_CPU_ARMV8 option.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Until now the "Target Architecture Variant" choice was not visible on
AArch64. In order to prepare the addition of the 64 bits core to this
choice, this commit adds a "depends on !BR2_ARCH_IS_64" dependency to
all currently supported cores (that are 32 bits only).
Following this commit, the "Target Architecture Variant" choice appears
on AArch64, but is for now empty.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The 64 bits ARM processors are capable of running 32 bits ARM code, and
some platforms are indeed using this capability. Due to this, if we were
to keep the separation between Config.in.aarch64 and Config.in.arm, we
would have to duplicate the definition of all 64-bits capable ARM cores
into both files.
Instead of going down this route, let's take the same route as the x86
one: a single Config.in.x86 file, used for both x86 32 bits and x86 64
bits, with the appropriate logic to only show the relevant cores
depending on which architecture is selected.
In order to do this, we:
- Make the "ARM instruction set" choice only visible on ARM 32 bits,
since we currently don't support ARM vs. Thumb on AArch64.
- Add the relevant values for the BR2_ARCH option.
- Add the relevant values for the BR2_ENDIAN option.
- Make the "aapcs-linux" BR2_GCC_TARGET_ABI value only used on ARM 32
bits, since this ABI doesn't mean anything on AArch64.
- Make the BR2_GCC_TARGET_FPU option depends on ARM 32 bits, since
there is no -mfpu option on AArch64.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This commit adds the option to select the Cortex-M4 ARM core, in the
same family as Cortex-M3. This will be useful to enable the internal
toolchain backend for this ARM core, and provide some defconfigs for
Cortex-M4 platforms.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
The Cortex-M cores only support Thumb-2, not Thumb. In fact, Thumb-2
is a superset of Thumb, and we could have a single option for both in
Buildroot, since -mthumb on ARMv4/v5 means original Thumb, while
-mthumb on ARMv7 means Thumb 2. However, for clarity, it makes sense
to have two separate options. But in this case, Cortex-M3 should not
advertise that it supports Thumb, as in fact selecting Thumb would
generate Thumb-2 code.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
All ARM cores should select a BR2_ARM_CPU_* option. Currently, the
cortex-m3 does not, which this commit fixes by introducing a
BR2_ARM_CPU_ARMV7M option.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Add the Cortex A17 variant. This core is considered a replacement
of the Cortex A12 and is supported by gcc 5 / binutils 2.25+
Suggested-by: Ross Green <greenfross@netscape.net>
Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Now that BR2_ARCH_HAS_ATOMICS is no longer used anywhere, we can
remove it from arch/Config.in*, as well as from the documentation.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
gcc differentiates the mpcore-with-vfp from the mcpore-without-vfp
CPUs. The former is named just 'mpcore', while the latter is named
'mpcorenovfp'.
We only add one entry, 'mpcore' and let the user select whether or
not to use the VFP. We then name the CPU according to the user's
selection.
Signed-off-by: Sergi Granell <xerpi.g.12@gmail.com>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
There's no point in offering the user an option to select an FP strategy
when the CPU does not actually have a VFP unit.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
There's no point in offering the user an option to select EABIhf when
the CPU does not really have a VFP unit.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Currently, the VFP selection for ARM is a little bit muddy:
- some CPUs definitely do not have a VFP or NEON,
- some CPUs definitely do have a VFP or NEON,
- some CPUs may have a VFP or NEON.
However, we currently conflate the availability of the VFP/NEON with the
possibility to use them. Even is the user chooses a floating point
strategy with a 'lower' solution (i.e. VFPv2 when a VFPv3 exists, or not
using NEON when the CPU has it), some packages are still using the
CPU-defined HW availaibility rather thean the usr's selection.
Furthermore, for CPU that may have a VFP/NEON, there is no way for the
user to actually specify that the HW is indeed available; the user can
only specify the floating point strategy. This means that some packages
or some package versions, like nodejs for example, can not be properly
selected on some CPU cores, like Cortex-A9 which only may have a VFP.
Like we have an option to enable an optional NEON unit, add a similar
option to enable an optional VFP unit.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Stating whether to use the NEON extensions when it is optional in the
CPU really is completing the definition of the CPU we've just selected.
Move the ENABLE_NEON option just after the choice of the CPU variant,
and before any "software" option (ABI/VFP).
This will make sense in a moment, when we introduce a similar option for
enabling an optional VFP unit.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
gcc will refuse to build with both --with-mode=thumb and --with-fpu=vfp,
with error messages during ./configure, like:
checking for suffix of object files... configure: error: in `/home/ymor
in/dev/buildroot/O/build/host-gcc-initial-4.9.3/build/arm-buildroot-lin
ux-uclibcgnueabihf/libgcc':
configure: error: cannot compute suffix of object files: cannot compile
See `config.log' for more details.
And config.log informatively contains:
sorry, unimplemented: Thumb-1 hard-float VFP ABI
This is an error message that comes deep from gcc source files.
If gcc says it does not support VFP with Thumb1, then let's disable that
combination in our menuconfig.
Prefer VFP over Thumb1, i.e. hide Thumb1 when we're not soft-float.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Set EABIhf as the default target ABI for the ARM processors that have or
may have a VFP unit, since this ABI is the most efficient in that case.
Of course, EABI can still be selected manually if needed.
[Peter: only default to EABIHF when we are sure the CPU has a VFP]
Signed-off-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Add the Cortex M3 variant. These microcontrollers don't support regular
ARM instructions and don't have an MMU.
Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Instead of blacklisting which architectures support MMUs (mandatorily
or optionally), introduce two Kconfig options that are selected by each
architecture in each case.
This simplifies the logic in BR2_USE_MMU.
Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Until now, all ARM processors supported the original ARM instructions.
However, the Cortex-M variants don't support them, and support only
Thumb/Thumb2 modes.
So, make a Kconfig option for ARM support and use it.
[Thomas:
- Remove the dependency in the choice between ARM/Thumb/Thumb-2,
because basically the choice is now always visible.
- Replace the BR2_ARM_INSTRUCTIONS_ARM_CHOICE choice option directly
by BR2_ARM_INSTRUCTIONS_ARM, instead of having this blind option
defined separately. This means the choice is now always visible,
even when only the ARM instruction set is supported.]
Signed-off-by: Guido Martínez <guido@vanguardiasur.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
On ARM, we were defining both the CPU type and the architecture
variant. However, depending on the version of gcc, a given combination
of (CPU, architecture) may not be the same. Since the architecture
variant is implied by the CPU type, given the former is not necessary,
and we can simply specify the latter.
>From the gcc documentation:
This specifies the name of the target ARM processor. GCC uses this
name to derive the name of the target ARM architecture (as if
specified by -march) and the ARM processor type for which to tune
for performance (as if specified by -mtune). Where this option is
used in conjunction with -march or -mtune, those options take
precedence over the appropriate part of this option.
Note that we verified that for all BR2_GCC_TARGET_ARCH value that
existed, a proper BR2_GCC_TARGET_CPU value is defined.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
In commit 88cf3bb917
("arch/Config.in.arm: Use armv6k for arm1136jf-s rev1"), Benoît
Thébaudeau added separate options for the revision 0 and revision 1 of
the ARM1136JF-S processor, so that different -march values could be
used (armv6j for revision 0, armv6k for revision 1).
However, this is preventing the removal of the BR2_GCC_TARGET_ARCH
option, which we need to do to give only the CPU type to gcc, and let
it decide the architecture variant that matches. This is because this
story of revision 0 vs. revision 1 is the only case where -mcpu
doesn't fully define the CPU.
Moreover, a quick test with gcc shows that -march=armv6j
-mcpu=arm1136jf-s is accepted, while -march=armv6k -mcpu=arm1136jf-s
makes gcc complain: " warning: switch -mcpu=arm1136jf-s conflicts with
-march=armv6k switch".
In addition, gcc 5 will apparently no longer allow to pass all of
--with-arch, --with-cpu and --with-tune, so we will anyway have to
rely only on one of them.
As a consequence, this commit basically reverts
88cf3bb917 and provides only one option
for ARM1136JF-S. If the two revisions are really different, then they
should be supported in upstream gcc with different -mcpu values.
Note that the removal of the two options should not break existing
full .config, since the hidden option BR2_arm1136jf_s becomes again a
visible option to select the CPU.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Benoît Thébaudeau <benoit.thebaudeau@advansee.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
In preparation to the removal of BR2_GCC_TARGET_ARCH for ARM, this
commit introduces a number of blind options for each ARM architecture,
so that packages/toolchains that had dependencies using
BR2_GCC_TARGET_ARCH can continue to express their dependencies. It can
also be used to simplify package dependencies that were using the
individual ARM core options.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
The BR2_arm10t option is not correct as it references an ARM family,
while other options indicate a specific ARM core. The ARM cores in
ARM10 family are ARM1020E, ARM1022E and ARM1026EJ-S according to
Wikipedia. However, those are clearly very rare, and Wikipedia only
indicates two Conexant ADSL-related SoC as being part of this family
of ARM cores. Therefore, this commit removes this ARM family.
[Peter: remove nettle.mk reference as pointed out by Yann]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The BR2_GCC_TARGET_CPU defines a value for the BR2_arm920 case, but
this option does not exist. Therefore, this commit removes one line of
dead code.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
armv6 and above all have one sort of atomic ops or another. For armv5
and below, they are emulated, either as a kernel trap, a kernel VDSO,
or compiler intrinsics.
Aarch64 is just armv8, so make it a single commit. ;-)
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Anton Kolesov <Anton.Kolesov@synopsys.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The toolchain currently doesn't build for nommu ARM and is in need of
serious work.
Problem is there are no emulation targets and real ARM(7TDMI/720T/740T)
hardware that's capable of running linux (enough memory, having a
memory controller...) is VERY rare and uses very old versions to
make it usable.
The ARM nommu focus should go into Cortex M series processors that are
obtainable at reasonable cost on modern hardware that has external
memory controllers.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Update the arm processor types: add the cortex A12 variant supported by
gcc 4.9.x
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Currently, the ARM Config.in logic specifies values for
--with-arch/-march and --with-tune/-mtune, but not for
--with-cpu/-mcpu. However, this causes problems on ARMv4, because
specifying --with-arch=armv4t isn't enough to make gcc generate ARMv4
code: one should also pass --with-cpu=<some ARMv4 CPU>.
Moreover, since Buildroot is generally designed to generate code
specifically for the configured target, it makes sense to give our own
--with-cpu/-mcpu value instead of relying on the default value used by
gcc, and only do small optimizations with -mtune.
Reported-by: Adam Hussein <kryme76@yahoo.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Commit 6b3a0417c4 ('arch/arm: arm926 may have VFP') forgot to update
the help text of the VFPv2 option to mention ARMv5. This commit fixes
that.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Until now, we were using the default ARM instruction set, as used by
the toolchain: the 32 bits ARM instruction set for the internal
backend, and for external toolchain, whatever default was chosen when
the toolchain was generated.
This commit adds support for the Thumb2 instruction set. To do so, it:
* provides a menuconfig choice between ARM and Thumb2. The choice is
only shown when Thumb2 is supported, i.e on ARMv7-A CPUs.
* passes the --with-mode={arm,thumb} option when building gcc in the
internal backend. This tells the compiler which type of
instructions it should generate.
* passes the m{arm,thumb} option in the external toolchain
wrapper. ARM and Thumb2 code can freely be mixed together, so the
fact that the C library has been built either ARM or Thumb2 and
that the rest of the code is built Thumb2 or ARM is not a problem.
[Peter: fix empty BR2_GCC_TARGET_MODE check]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
This commit introduces the support for the EABIhf ABI, next to the
existing support we have for EABI and OABI (even though OABI support
is deprecated). EABIhf allows to improve performance of floating point
workload by using floating point registers to transfer floating point
arguments when calling functions, instead of using integer registers
to do, as is done in the 'softfp' floating point model of EABI.
In addition to this, this commit introduces a list of options for the
floating point support:
* Software floating point
* VFP
* VFPv3
* VFPv3-D16
* VFPv4
* VFPv4-D16
and it introduces some logic to make sure the options are only visible
when it makes sense, depending on the ARM core being selected. This is
however made complicated by the fact that certain VFP capabilities are
mandatory on some cores, but optional on some other cores. The kconfig
logic tries to achieve the following goals:
* Hide options that are definitely not possible.
* Use safe default values (i.e for Cortex-A5 and A7, the presence of
the VFPv4 unit is optional, so we default on software floating
point on these cores)..
* Show the available possibilities, even if some of them are not
necessarily working on a particular core (again, for the Cortex-A5
and A7 cores, there is no way of knowing whether the particular
variant used by the user has VFPv4 or not, so we select software
floating point by default, but still show VFP/VFPv3/VFPv4 options).
It is worth noting that this commit doesn't add support for all
possible -mfpu= values on ARM. We haven't added support for fpa, fpe2,
fpe3, maverick (those four are only used on very old ARM cores), for
vfpv3-fp16, vfpv3-d16-fp16, vfpv3xd, vfpv3xd-fp16, neon-fp16,
vfpv4-sp-d16. They can be added quite easily if needed thanks to the
new organization of the Config.in options.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
As we are going to introduced a more advanced support of floating
point options for the ARM architecture, we need to adjust how the
soft-float option is handled. We replace the current hidden option
BR2_PREFER_SOFT_FLOAT option and the visible BR2_SOFT_FLOAT option by:
* A global hidden BR2_SOFT_FLOAT option, defined in arch/Config.in,
that tells whether the architecture-specific code is using software
emulated floating point. This hidden option can be used throughout
Buildroot to determine whether soft float is used or not.
* Per-architecture visible BR2_<arch>_SOFT_FLOAT options, for the
architecture for which it makes sense, which allows users to select
soft float emulation when needed.
This change will allow each architecture to have a different way of
presenting its floating point capabilities.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
OABI is more than legacy, it's dead.
New developments should go with EABI, since it so much better.
>From the Debian EABI page [0] :
- floating point performance, with or without an FPU is very much faster
- mixing soft and hardfloat code is possible
- structure packing is not as painful as it used to be
- a more efficient syscall convention
- more compatibility with various tools
[0] http://wiki.debian.org/ArmEabiPort
[Thomas: keep the ABI choice, as we are going to introduce EABIhf later].
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
