Adds a new compiler flag to enable function instrumentation injection
at compile time for compilers that support it.
Adds support for this flag to GCC when `CONFIG_INSTRUMENTATION` is
enabled, setting the `-finstrument-functions` flag during compilation.
Signed-off-by: Kevin Townsend <kevin.townsend@linaro.org>
Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Signed-off-by: Maciej Sobkowski <msobkowski@antmicro.com>
Fixes#91045
Adding NO_SPLIT in the TOOLCHAIN_LD_FLAGS list caused every flags in the
list to be checked together by the linker. This caused issues where
incompatible flags was passed together, and the result was that none was
propagated to the linker.
By using separate variables, we can achieve the initial goal of NO_SPLIT to
only group a sub-set of flags.
Signed-off-by: Pierrick Guillaume <pguillaume@fymyte.com>
Compiler property no_track_macro_expansion is controlled by
CONFIG_COMPILER_TRACK_MACRO_EXPANSION.
When that Kconfig is set to Y, clang fails with the following error
message:
unknown argument: '-ftrack-macro-expansion=0'
This commit modifies clang/compiler_flags.cmake so that there is a
proper clang compiler option for this flag.
Signed-off-by: Benjamin Cabé <benjamin@zephyrproject.org>
-Wno-volatile is set by gcc/compiler_flags.cmake,
but is not supported by clang and generated warnings.
Signed-off-by: Jaagup Averin <jaagup.averin@gmail.com>
Since this overrides values from gcc/compiler_flags.cmake,
they should ideally have the same structure and order.
Signed-off-by: Jaagup Averin <jaagup.averin@gmail.com>
Provide toolchain-specific implementations for the new functions
added in d77b58a, because the default ones end up with build failures.
As the MWDT toolchain doesn't support different combinations of
compilers and linkers, there's no need to pass any linker properties
from the compiler, at least for the time being. Moreover, the default
implementation of compiler_set_linker_properties() uses the compiler
flag (--print-libgcc-file-name) that doesn't make any sense for MWDT.
Therefore, provide an empty implementation of the function.
The default implementation of toolchain_linker_add_compiler_options()
doesn't cause any issues for now, but it's still better to replace
it with the passthrough implementation borrowed from ld/lld.
Signed-off-by: Ilya Tagunov <Ilya.Tagunov@synopsys.com>
Use CMAKE_C_COMPILER_TARGET to allow clang to correctly determine
the runtime library for the target architecture.
Otherwise, the default runtime library will be selected.
Signed-off-by: Mikhail Siomin <victorovich.01@mail.ru>
With inclusion of the optimization flag into the multilib selection
process, we cannot compute the compiler library path when the compiler's
target.cmake is processed as LINKER_OPTIMIZATION_FLAG is not computed until
much later.
Instead, add a function (compiler_file_path) which can be used to locate
the appropriate crtbegin.o and crtend.o files.
Delay computation of lib_include_dir and rt_library until after all
compiler flags have been computed by adding compiler_set_linker_properties
and calling that just before toolchain_linker_finalize is invoked.
Place default implementations of both of these functions in a new file,
cmake/compiler/target_template.cmake, where we assume the compiler works
like gcc or clang and handlers the --print-file-name and
--print-libgcc-file-name options. Compilers needing alternate
implementations can override these functions in their target.cmake files.
These implementations require that no generator expressions are necessary
for the compiler to compute the right library paths.
This mechanism is also used to take any additional compiler options by
adding a new (optional) linker function,
toolchain_linker_add_compiler_options, which maps compiler options to
equivalent linker options, discarding any that aren't applicable.
Signed-off-by: Keith Packard <keithp@keithp.com>
Commit f499ea0 made it so that the compiler optimization flags are also
passed to the assembler. However, this breaks for the IAR toolchain
since it uses the GNU assembler which does not have the same
optimization flags as the IAR compiler.
This commit separates the compiler and assembler optimizations to make
it possible to specify appropriate flags in a toolchains corresponding
`compiler_flags.cmake`
Signed-off-by: Daniel Flodin <daniel.flodin@iar.com>
Allow users to compile LLEXTs with the MetaWare Design Toolchain.
Provides utility to dramatically shrink bloated extensions by removing
unused section names left behind by MWDT strip. Test extensions will
not fit in NSIM VPX5 memory otherwise; each is reduced from ~16k to ~1k.
Signed-off-by: Lauren Murphy <lauren.murphy@intel.com>
to be able to choose c++23 dialect
deprecates c++2b, but we should keep the option c++2b for legacy
Signed-off-by: Adrian Bieri <adrian.bieri@loepfe.com>
This commit introduces a new Kconfig choice for configuring the code model
used for compilation.
All three code models specified by the RISC-V ELF psABI specification are
supported: medlow, medany and large.
For RV32, the `medlow` code model is always selected.
For RV64, the `large` code model is selected if the SRAM or kernel VM base
address is greater than or equal to 4 GiB; otherwise, the `medany` code
model is selected.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit introduces the support for the standard Zicntr extension, which
provides hardware counters.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds the support for the following new Zc* series compressed
instruction extensions:
Zca - Subset of the C extension without FP loads and stores
Zcb - "Simple" instructions
Zcd - Double-precision floating-point instructions
Zcf - Single-precision floating-point instructions
Zcmp - "Complex" instructions for embedded CPUs
Zcmt - Table jump instructions for embedded CPUs
With the introduction of the Zc* extensions, the C extension now implies
the following Zc* extensions:
* Zca, always
* Zcf if F is specified (RV32 only)
* Zcd if D is specified
The Zc* extensions that are implied by the C extension are not specified in
the GCC `-march` flag because they are redundant and can interfere with the
resolution of the correct multi-lib for the selected architecture unless
the the alternate mappings for the redundant forms are manually specified.
All the implementation details in this commit are based on the Zc* v1.0.0
specification, which is the ratified version.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Fixed a problem where the command line option --vla was used
with C++ causing a command line error in the IAR ICCARM compiler.
Signed-off-by: Lars-Ove Karlsson <lars-ove.karlsson@iar.com>
This commit fixes the incorrect code formatting used by target_riscv.cmake
by changing the 4-space indent to 2-space and removing the space between if
keyword and parenthesis.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Several years ago we started enabling the -Waddress-of-packed-member
warning in commit cf111962e0 for GCC.
Do the same for Clang, since there is no reason not to and the warning
is very useful, especially in protocol stacks.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Commit be40d854c2ccacf14ca3fcfb01bffdc9b075c6c9 introduced the ability
of building Zephyr with deprecation warnings enabled, by making
COMPILER_WARNINGS_AS_ERRORS depend on the newly added DEPRECATION_TEST
Kconfig option. This has the downside of disabling **all** warnings, not
only the deprecation ones.
This patch instead makes DEPRECATION_TEST disable only the deprecation
warning, but leaves COMPILER_WARNINGS_AS_ERRORS enabled.
This has the advantage of being able to see other unrelated warnings
(and fail if they appear) but has the disadvantage of not printing out
the deprecation warnings themselves (since they are disabled).
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Although xt-clang is based on clang, for some reason, it still
lists xcc system include path as the first search path (e.g.
for stddef.h), and the clang system include path as last. This
creates a big issue when the code starts to use any standards
past C89 (since xcc is based on GCC 4.2). We can use compiler
property nostdin_include to add -isystem to compiler options.
However, some modules (e.g. picolibcs) somehow ignore this.
So we also need to forcibly do add_compile_options() to make
sure the clang system include path is placed before the xcc
system include path.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Xtensa toolchain xt-clang needs similar treatment to xcc in
terms of nostdinc and nostdinc_include, so add it to the if
block to exclude these by default.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
"-fshort-enums" changes the ABI. It's not enabled for gcc, so it's not
clear why it's enabled for clang (only ARM) and armclang, other than it
looks like some users of these toolchains may be linking against code
that is compiled with "-fshort-enums".
As an example, when compiling with clang, CONFIG_LTO, and a toolchain
built without "-fshort-enums", the linker warns:
ld.lld: error: linking module flags 'min_enum_size': IDs have
conflicting values in
'/usr/armv7m-cros-eabi/usr/lib/libc++_static.a(string.cpp.o at 784090)'
and 'ld-temp.o'
Signed-off-by: Tom Hughes <tomhughes@chromium.org>
When running naive sim builds using clang, respect user passed
SYSROOT_DIR values. Additionally, allow overriding the archive tool
when calling Make for the native simulator.
Add the sysroot (if available) to the native_simulator target and
TOOLCHAIN_C_FLAGS and TOOLCHAIN_LD_FLAGS. Additionally pass CMAKE_AR
to NSI_AR.
Signed-off-by: Yuval Peress <peress@google.com>
Rework how the compilation-options for the gperf generated
kobject_hash*.c files are put together to fix problems with SHELL:
and cmake-list separators handling.
zephyr_get_compile_options_for_lang_as_string() returns the set of
options as a cmake generator expression string which is cumbersome to
edit. This caused the command line for the IAR toolchain to have broken
SHELL: entries, and other some command line entries being postfixed by
"gnu".
This also adds CMake compiler properties for no_function_sections and
no_data_sections options
Signed-off-by: Björn Bergman <bjorn.bergman@iar.com>
The RX floating-point hardware only works on 32-bit values
so default gcc will build doubles as 32-bit, this make many
test fail on RX build test.
This commit force the build for RX arch to use 64-bit doubles
FPU will be supported later on this arch
Signed-off-by: Duy Nguyen <duy.nguyen.xa@renesas.com>
The QingKe V2C has an integer multiplier but no divide. Add support
for the corresponding Zmmul extension and, as the extension was added
in GCC 13.0, add a test for the compiler version.
Signed-off-by: Michael Hope <michaelh@juju.nz>
GCC and Clang support the undefined behavior sanitizer in any
configuration, the only restriction is that if you want to get nice
messages printed, then you need the ubsan library routines which are only
present for posix architecture or when using picolibc.
This patch adds three new compiler properties:
* sanitizer_undefined. Enables the undefined behavior sanitizer.
* sanitizer_undefined_library. Calls ubsan library routines on fault.
* sanitizer_undefined_trap. Invokes __builtin_trap() on fault.
Overhead for using the trapping sanitizer is fairly low and should be
considered for use in CI once all of the undefined behavior faults in
Zephyr are fixed.
Signed-off-by: Keith Packard <keithp@keithp.com>
Removes functions that would handle deprecated variables and
functionality from CMake within Zephyr that were deprecated in
Zephyr 3.1 onwards but before 3.7
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
Rewrote the warning levels for toolchain IAR as IAR tools just turn
off warnings, not on. Also did some minor cleanup for coding
guidelines.
Signed-off-by: Lars-Ove Karlsson <lars-ove.karlsson@iar.com>
xt-clang is usually based on older version of clang, and
Zephyr main targets more recent versions. Because of this,
some newer compiler flags may cause warnings where twister
would mark as test being failed. To workaround that,
add -Wno-unknown-warning-option to suppress those warnings.
Fixes#84138
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Introduce GCC_COMPILER_VERSION to be used in Zephyr CMake when knowledge
of the current GCC Compiler version is needed and
CMAKE_C_COMPILER_VERSION has not been defined yet.
CMAKE_C_COMPILER_VERSION is not defined until project() is called.
Signed-off-by: Torsten Rasmussen <torsten.rasmussen@nordicsemi.no>
CMSIS modules dsp and nn sets the optimization flag -Ofast for their
libraries. I've now made a new compiler property optimization_fast to
avoid using ifdefs in the cmsis CMakeLists.txt files.
I've update GCC and arcmwdt (not sure if arcmwdt supports -Ofast)
Signed-off-by: Lars-Ove Karlsson <lars-ove.karlsson@iar.com>
Currently some subsystems outside the zephyr kernel
use VLA:s such as bluetooth. We will enable it for now
in the same kind of situations as gcc.
Signed-off-by: Robin Kastberg <robin.kastberg@iar.com>
On RISC-V, GCC and possibly other compilers include small globals in
special "small data" sections like .sbss and .sdata.
This allows accessing these globals using the global pointer register
and an offset, which is faster than ordinary relocations.
Currently, llext_load does not support multiple NOBIT sections.
Thus, llext's that use both .bss and .sbss sections cannot be relocated.
Also, the global pointer cannot be used in llext's, as it does not
change when calling into or returning from an llext, making the
optimization pointless for llext.
This commit disables the use of small data sections by adding the
appropriate compiler flag when compiling llext's.
In addition to solving the aforementioned issue, this fixes llext's
test_inspect on RISC-V.
Signed-off-by: Eric Ackermann <eric.ackermann@cispa.de>
Add the -imacros property for assembler language to compensate
for changes introduced in PR #84800.
Signed-off-by: Björn Bergman <bjorn.bergman@iar.com>
The IAR linker does not handle Dwarf 5 and since it uses the GCC
assembler for assembler files, we need to make sure the assembler
output uses Dwarf 4.
Signed-off-by: Lars-Ove Karlsson <lars-ove.karlsson@iar.com>
Some basic flags that were hardcoded needed
to be moved to properties to be able to implement
alternative toolchains.
Signed-off-by: Robin Kastberg <robin.kastberg@iar.com>
