This commit adds a template for specifying the C/C++ compiler flag for
disabling the strict aliasing rule.
It also enables this flag globally because the Zephyr codebase does not
strictly adhere to the aliasing rules specified by the C/C++ standards
and the optimisation strategies that assume these rules may end up
generating invalid code.
For instance, GCC 11 and above tend to optimise more aggressively
assuming the strict adherence to the standard aliasing rules and may
generate invalid code, when compiling Zephyr with `-fstrict-aliasing`,
that results in various run-time failures.
Note that the footprint and performance ramifications of disabling the
strict aliasing rule are negligible.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Devicetree fixup files existed previous to the current stable Devicetree
API. While they served their purpose, they are no longer necessary nor
used in-tree. This patch drops support for this legacy feature.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
dts.fixup was a thing before dts_fixup.h existed. Remove the check as we
have already moved away from dts_fixup.h.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Output the final dependency graph as a `.dot` file, which when rendered
by graphviz can be easier to comprehend than the text descriptions.
This output is optional in that it will not be generated if `graphviz`
is not installed.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
SoCs can have different headers, make the build system include those if
the directory exists for this SoC.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
With support for deprecating Kconfig symbols within the Kconfig tree
itself, there is no reason to have a secondary location to maintain
for information that is based on a Kconfig setting in first place.
Better to directly do `select DEPRECATED` on the symbol itself, than to
maintain a secondary list with messages in CMake.
Hence removing this feature in CMake.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Move scripts needed by the build system and not designed to be run
individually or standalone into the build subfolder.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Optional way for generating symbols file using binutils. This file will
be generated on demand for tests which will then be parsed by twister to
create an accurate test plan from the listed testcase/suites.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
There is no reason to have this script in a different place than all the
other python scripts. Move it.
Signed-off-by: Carlo Caione <ccaione@baylibre.com>
Compilation warnings appears for C++ files, that following
options are not valid:
-ffrestanding,
-Wno-format-zero-length
-Wno-main
-fgnu89-inline
-std-gnu99
Added checks to filter out unsupported flags.
Signed-off-by: Jaroslaw Stelter <Jaroslaw.Stelter@intel.com>
Picolibc is a fork of newlib designed and tested on embedded systems. It
offers a smaller memory footprint (both ROM and RAM), and native TLS
support, which uses the Zephyr TLS support.
By default, the full printf version is included in the executable, which
includes exact floating point and long long input and output. A
configuration option has been added to switch to the integer-only
version (which also omits long long support).
Here are some size comparisons using qemu-cortex-m3 and this application
(parameters passed to printf to avoid GCC optimizing it into puts):
void main(void)
{
printf("Hello World! %s %d\n", CONFIG_BOARD, 12);
}
FLASH SRAM
minimal 8696 3952
picolibc int 7600 3960
picolibc float 12304 3960
newlib-nano int 11696 4128
newlib-nano float 30516 4496
newlib 34800 6112
---
v2:
Include picolibc-tls.ld
v3:
Document usage in guides/c_library.rst and
getting_started/toolchain_other_x_compilers.rst
v4:
Lost the lib/libc/picolibc directory somehow!
v5:
Add PICOLIBC_ALIGNED_HEAP_SIZE configuration option.
Delete PICOLIBC_SEMIHOST option support code
v6:
Don't allocate static RAM for TLS values; TLS
values only need to be allocated for each thread.
v7:
Use arm coprocessor for TLS pointer storage where supported for
compatibility with the -mtp=cp15 compiler option (or when the
target cpu type selects this option)
Add a bunch of tests
Round TLS segment up to stack alignment so that overall stack
remains correctly aligned
Add aarch64 support
Rebase to upstream head
v8:
Share NEWLIB, NEWLIB_NANO and PICOLIBC library configuration
variables in a single LIBC_PARTITIONS variable instead of
having separate PICOLIBC_PART and NEWLIB_PART variables.
v9:
Update docs to reference pending sdk-ng support for picolibc
v10:
Support memory protection by creating a partition for
picolibc shared data and any pre-defined picolibc heap.
v11:
Fix formatting in arch/arm/core/aarch64/switch.S
v12:
Remove TLS support from this patch now that TLS is upstream
Require THREAD_LOCAL_STORAGE when using PICOLIBC for architectures
that support it.
v13:
Merge errno changes as they're only needed for picolibc.
Adapt cmake changes suggested by Torsten Tejlmand Rasmussen
v14:
Update to picolibc 1.7 and newer (new stdin/stdout/stderr ABI)
v15:
Respond to comments from dcpleung:
* switch kernel/errno to use CONFIG_LIBC_ERRNO instead of
CONFIG_PICOLIBC
* Add comment to test/lib/sprintf as to why the %n test
was disabled for picolibc.
v16:
Switch picolibc to a module built with Zephyr. This eliminates
toolchain dependencies and allows compiler settings for Zephyr
to also be applied to picolibc.
v17:
Provide Zephyr-specific 'abort' implementation.
Support systems with MMU
v18:
Allow use of toolchain picolibc version.
v19:
Use zephyr/ for zephyr headers
v20:
Add locking
Use explicit commit for picolibc module
v21:
Create PICOLIBC_SUPPORTED config param. Set on arc, arm, arm64,
mips and riscv architectures.
Signed-off-by: Keith Packard <keithp@keithp.com>
There are few cases where removing this enforcement is desirable:
- linking against binary C++ libs with incompatible type mangling
- linking against system provided headers i.e. native_posix
- compiling with legacy code that assumes a different convention
So let's create a Kconfig symbol for it. This is IMHO a good compromize
compared to using the %"PRId32" abomination everywhere otherwise.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Avoid potentially calling __builtin_clz() twice with non-constant
values. Also add a test for it.
Clang produces false positive vla warnings so disable them. GCC will
spot real vla's already.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
We currently have mechanisms for deprecating both boards and SoCs.
However, we lack one for deprecating modules. This is inconvenient,
because we would like to do exactly that.
Handle this in a simple way by adding a new CMake file in the modules
directory which is responsible for warning the user about any
deprecated modules they may be using. See the source code comments for
more details about the approach.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
We want to use a sparse address space to identify invalid conversions
between cached and uncached address aliases. This patch adds a
__sparse_cache sparse annotation for that. Where those conversions
must be done that has to be supported by using the __sparse_force
sparse attribute. To avoid compiler complains about unknown
attributes we add a -Wno-attributes flag when building with sparse
support.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
With this adding "-DSPARSE=y" to the "west build" command line
performs a sparse check of the project build. So far only gcc-based
builds are supported.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Add API to add devices to a power domain in runtime. The number of
devices that can be added is defined in build time.
The script gen_handles.py will check the number defined in
`CONFIG_PM_DEVICE_POWER_DOMAIN_DYNAMIC` to resize the handles vector,
adding empty slots in the supported sector to be used later.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Cleanup in kconfig options in preparation for adding a
frontend that will use dictionary mode.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Add a Kconfig to disable non prefixed includes. By setting
`CONFIG_LEGACY_INCLUDE_PATH=n` developers can disable having
`include/zephyr` in the search path. This means that includes such
as `#include <kernel.h>` will no longer work.
Internally, every header should be updated to add the `zephyr/`
prefix to the headers. Only then, will developers be able to use
this config value for their applications.
Signed-off-by: Yuval Peress <peress@google.com>
Move include paths and add new target_include_directories to support
backwards compatibility:
* /include -> /include/zephyr
example: <irq.h> -> <zephyr/irq.h>
Issue #41543
Signed-off-by: Yuval Peress <peress@google.com>
MIPI Sys-T catalog messages are similar to dictionary logging
where an ID is emitted instead of the format string. This allows
the format strings to be removed from the final binary to save
a few bytes. This adds the necessary bits to determine to emit
catalog messages when appropriate.
Note that this implementation copies the argument list as-is
with string arguments stitched together since the format strings
are assumed to have been removed and they cannot be examined
to properly convert the argument lists into catalog message
payloads. Because of this, various build asserts are there to
avoid building for configurations where they are known not to
work.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This changes the script to allow output format to be specified.
Currently, only JSON is support. This will allow supporting
other formats in the future.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
In cmake we use target_byproducts() to register dependence on the map
file alongside the executable for it - this makes sense.
Thus if the map file is missing, ninja will detect and issue a linker
command again. However after that, cmake was instructed to rename the
map file. Thus a 2nd round of ninja, which should be a no-op, is not as
the registered byproduct file is missing.
To keep this static, and to keep the map file alongside the elf file, we
will instead copy the map file.
Signed-off-by: Mark Ruvald Pedersen <mped@oticon.com>
In 9170977 build-time version header generation was added. The test
for .git assumes this file to be a directory. In the case of git
submodules, .git is a regular file that in its contents points to
the actual git database for the submodule. This is a way to have
symlink like behaviour even on file systems that do not support
themselves support symlinks.
This consults git as to what the correct git database directory is,
in case the .git file is indeed a regular file, and adjusts the
git_dependency variable accordingly.
Fixes#43503
Signed-off-by: Frank Terbeck <ft@bewatermyfriend.org>
Add symbols from libm.a or libm_nano.a to z_libc_partition. This fixes
an issue where newlib math functions called from user mode thread would
cause an MPU fault.
Fixes#43661
Signed-off-by: Helge Juul <helge@fastmail.com>
.eh_frame section should not be removed directly in the
hex format and bin format output, it should be based on
whether we need exception handler feature.
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
With CONFIG_TIMEOUT_64BIT it is both k_timeout_t and k_ticks_t that
need to be split, otherwise many syscalls returning a number of ticks
are being truncated to 32 bits.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Some emulation definition might be defined out of tree. Do not try to
include them here, instead they will be included in the module tree
defining them.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
As part of #40167 is was discovered that devicetree headers are always
generated when CMake re-runs.
This causes all source files that directly or indirectly through
included headers to recompile when CMake re-runs, even if there are no
changes to devicetree.
This commits introduces `zephyr_file_copy(...)` similar to
`file(COPY_FILE ...)` from CMake 3.21.
However, as CMake 3.20 is supported by Zephyr we need a zephyr variant
of this function to allow usage with CMake 3.20.
This ensures that only when there are changes to devicetree headers,
then source files will recompile.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
Fixes: #39503Fixes: #40167
This commit moves `BUILD_VERSION` CMake variable from a compile
definition to be a define inside version.h.
Besides the benefit of having related settings grouped in a common
header, it also means that an updated `BUILD_VERSION` value does not
need to trigger re-compilation of all source files.
When using compile definitions, CMake cannot tell whether a given source
files uses the definition or not, and hence all sources must be
recompiled to be sure they are up-to-date.
Placing `BUILD_VERSION` in version.h, the source dependencies ensures
that only source files including `version.h` gets recompiled.
As part of this, version.h generation is moved so that it is now done
at build time.
This means that re-generation of version.h is no longer depending on a
CMake re-run but can have it's own dependency in `.git/index` when git
described is used to obtain `BUILD_VERSION` information.
Generation of logging dictionary database has been updated to support
BUILD_VERSION from header file instead of CMake configure time variable.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
Fixes: #42184
This commit fixes dependency issues related to kobject hash generation.
Absolute path is ensured in cases where OUTPUT was provided with
absolute path, as `${CMAKE_CURRENT_BINARY_DIR}`.
DEPENDS referring to same file has been updated to also use
`${CMAKE_CURRENT_BINARY_DIR}` to ensure they reference identical
locations.
The custom command renaming sections in kobject_hash.o and creating
kobject_hash_renamed.o has been updated to depends on the target objects
of kobj_hash_output_lib in addition to the library itself.
This is needed because kobj_hash_output_lib is not a real library, and
thus no library is updated when this target rebuilds.
Instead the dependency must be on the object files created by
kobj_hash_outplut_lib. This ensures that when object files gets rebuilt
then the section renaming will also take place.
The reason why both the object library itself, and its object files are
required as dependencies has to do with build generators.
The library is needed to ensure Makefiles can correctly have a target to
invoke when the output of the custom command is missing. The object
files dependency is required to ensure that custom commands are
correctly brought up-to-date when the objects changes.
Similar, the custom command executing gen_kobject_placeholders.py
depending of kobj_prebuilt_hash_output_lib has been updated to also
depend on the object files created by kobj_prebuilt_hash_output_lib.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
UF2 offset configurable to be either flash base reg + offset,
or simply the flash bare reg if the offset is configured
to be ignored.
Signed-off-by: Peter Johanson <peter@peterjohanson.com>
This commit adds support for adjust the addresses of the final image.
This is useful when the image is to be flashed at a location different
from the LMA address encoded in the ELF file by the linker.
An example use-case is multicore systems where core A might load image
from a flash partition into RAM in order for core B to execute and load,
but where the image itself is build with the RAM addresses as LMA.
It updates the zephyr_image_info.h header with information of adjustment
value.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
This commit adds the `gen_image_info.py` script which supports creation
of a header file with image information from the EFL file.
This version populates the header file with:
- Number of segments in the image
- LMA address of each segment
- VMA address of each segment
- Size of each segment
The header file can be used by a secondary build system which needs this
information.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
The `-fno-strict-overflow` option blocks some gcc optimizations,
removing it in SOF showed a measurable performance improvement: see code
review #41457 of commit 60325019aa ("samples/subsys/audio/sof: use
-fstrict-overflow for SOF").
Add these optimizations to every Zephyr project beyond the somewhat
awkward and maybe temporary way the SOF `samples/` (?) is currently
built. There are non-SOF specific discussions in #41457 BTW.
Fixes#8924 which already had a plan to remove this option in 2018. It
looks like some bad Github feature unfortunately auto-closed the issue
before it was actually removed.
`-fno-strict-overflow` is needed only for code that relies on signed or
pointer wrapping, which is an undefined C behavior. Most security
policies seem to forbid undefined C behaviors.
(Digression: _unsigned_ integer wrapping is _defined_ behavior and
expected from any half-decent compiler.)
Until gcc version 7, the default -fstrict-overflow value was depending
on the -On optimization level. In gcc version 8, the whole feature was
greatly simplified: -fnostrict-overflow became a simple alias for
`-fwrapv` and `-fwrapv-pointer`: wrapping or no wrapping, period. For
the subtle, pre-v8 difference between -fno-strict-overflow and -fwrapv
please read this great, pre-v8 intro from the author of strict-overflow:
- https://www.airs.com/blog/archives/120
And also:
- https://gcc.gnu.org/onlinedocs/gcc-11.2.0/gcc/Code-Gen-Options.html
- https://gcc.gnu.org/onlinedocs/gcc-7.3.0/gcc/Optimize-Options.html
Important quote from Code-Gen-Options.html:
> Most of the options have both positive and negative forms; the
> negative form of -ffoo is -fno-foo. In the table below, only one of the
> forms is listed: the one that is not the default.
This means _undefined_ wrapping is the default. So simply removing this
zephyr_cc_option() line is enough to switch and great if any project
desires overriding it locally: no need to scrutinize the command line
order and pray it does the right thing (have a look at the precedence
between -f[no]-wrapv and -f[no-]trapv for some of that fun).
With gcc, the documented `-Wall` list includes `-Wstrict-overflow=1`:
- https://gcc.gnu.org/onlinedocs/gcc-11.2.0/gcc/Warning-Options.html
-Wextra does not increase the level of this warning.
This warning reports some of the cases where a gcc optimization comes
and actually "breaks" signed (and undefined!) wrapping. A real example
was just fixed in commit f38c6b67ed ("samples: big_http_download: make
num_iterations unsigned"). Note this reporting is incomplete and it also
assumes that developers look at warnings (not always true) or that CI
defines `-DEXTRA_CFLAGS=-Werror` as it should (I am pretty sure I
remember Zephyr CI adding -Werror in the past but I'm afraid it's gone
right now).
Increasing the level to -Wstrict-overflow=2 or more seems to report too
many false positives: #41551
Additional info:
- gcc also supports `-ftrapv` to catch signed overflow at _run-time_.
- Clang has `-fsanitize=signed-integer-overflow`. I did not test it.
https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html
- zephyr_cc_option() silently discards its argument for toolchains that
fail it (so arguments with typos are silently ignored globally)
Signed-off-by: Marc Herbert <marc.herbert@intel.com>
Fixes: #41435
Remove build host specific escaping of start symbol command argument.
The start symbol for armlink is: Image$$device$$Base
and were escaped as: Image\\$$\\$$device\\$$\\$$Base
However, the $ must only be escaped in Linux and MacOS, not on windows
hosts.
Instead of escaping the start symbol in the CMake code then it is better
to use the VERBATIM flag on `add_custom_command()` which ensures correct
escaping for the build host.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
We used to test that the compiler works at all, before testing
compiler flags. But during some toolchain refactoring this got
re-arranged.
This commit corrects the ordering.
Signed-off-by: Sebastian Bøe <sebastian.boe@nordicsemi.no>
