The bootloader subproject's full Kconfig discovery resolves CONFIG_SPIRAM=y
when the parent app has it enabled, even though esp_psram is not linked
into the bootloader (the CMake gate is
'if(NOT non_os_build) if(CONFIG_SPIRAM) idf_component_optional_requires(PRIVATE esp_psram)').
Shared sources in esp_hw_support that #include esp_psram private headers
or call esp_psram functions guarded only by '#if CONFIG_SPIRAM' then fail
to compile in the bootloader with
"fatal error: esp_private/esp_psram_extram.h: No such file or directory".
Mirror the CMake gate in source guards: every '#if CONFIG_SPIRAM' block in
a bootloader-compiled source that touches esp_psram becomes
'#if !BOOTLOADER_BUILD && CONFIG_SPIRAM'. The leaked CONFIG_SPIRAM value
in the bootloader's sdkconfig.h is then harmless because every dependent
block evaluates to false.
Sites updated:
- esp_memory_utils.c: include of esp_psram_extram.h and all
esp_psram_check_ptr_addr() call sites
- port/esp32{c5,c61,p4}/cpu_region_protect.c: include of
esp_psram_extram.h (inner SPIRAM_FETCH/RODATA/PRE_CONFIGURE blocks are
already inside outer !BOOTLOADER_BUILD guards)
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
When a component's manifest declared a managed dep `<ns>/<name>` and the
project had a local `components/<name>` shadowing it, configure failed
with `IDF: Failed to resolve component '<ns>__<name>'`.
cmakev1 does a project-wide injection so the manager's _choose_component
sees every component name in known_components and rewrites the
namespaced dep to its locally-shadowing short name. cmakev2 invokes the
manager per component with a single-component requirements file, so
known_components has only one entry and the rewrite cannot fire --
`<ns>__<name>` is written verbatim into MANAGED_REQUIRES and the
force-include at component.cmake:995 aborts.
Seed the per-component requirements file with one __COMPONENT_SOURCE
entry per discovered component so handle_project_requirements() builds
the same known_components list cmakev1 produces. The seeded entries are
inert on the cmake side -- the __component_set_property shim ignores
__COMPONENT_SOURCE -- they exist purely to feed the manager's rewrite.
Add a regression test in tools/test_build_system/buildv2/test_component.py.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
When CONFIG_SECURE_BOOT_V2_ENABLED=y but
CONFIG_SECURE_BOOT_BUILD_SIGNED_BINARIES is not set, produce the
binary directly as bootloader.bin instead of bootloader-unsigned.bin.
This matches the v1 behavior where the intermediate binary name is
conditional: bootloader-unsigned.bin only when build-time signing is
enabled (so the signed output can be named bootloader.bin), otherwise
the output is bootloader.bin directly.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The __init_project_configuration() function in cmakev2's project.cmake
unconditionally applied app-level compiler optimization flags based on
CONFIG_COMPILER_OPTIMIZATION_* Kconfig options. When the bootloader
subproject was built with cmakev2, these app-level flags leaked into the
bootloader compile command alongside the correct bootloader-specific
flags from CONFIG_BOOTLOADER_COMPILER_OPTIMIZATION_*.
For example, with the default configuration (app: DEBUG, bootloader:
SIZE), the bootloader received both "-Og -fno-shrink-wrap" (from app
config) and "-Os -freorder-blocks" (from bootloader config). While GCC
uses the last -O flag (-Os wins), the stray -fno-shrink-wrap persisted.
Introduce a SET_COMPILER_OPTIMIZATION build property that defaults to
YES when unset. Subprojects that manage their own optimization flags
(like the bootloader) can set this to NO before calling
idf_project_init() to prevent the default optimization flags from being
applied. This keeps project.cmake generic without requiring it to know
about specific subproject types.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The idf_path variable was used in -fmacro-prefix-map and
-fdebug-prefix-map flags but never read from the IDF_PATH build
property, resulting in an empty substitution. This caused full
filesystem paths to leak into .rodata instead of being mapped to /IDF.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The bootloader does not use the heap allocator or TLSF, so the ROM
patch files esp_rom_tlsf.c and esp_rom_multi_heap.c are not needed
in the bootloader build.
In the cmakev1 build system this was never an issue because cmakev1
uses an early expansion phase where only Kconfig files for components
listed in BUILD_COMPONENTS are processed. Since the heap component is
not part of the bootloader's component list, its Kconfig options
(CONFIG_HEAP_TLSF_USE_ROM_IMPL, etc.) were never defined and the
conditional compilation of these source files was effectively skipped.
In the cmakev2 build system, Kconfig options from all discovered
components are visible regardless of whether the component is part
of the build. Because the bootloader reuses the main project's
sdkconfig (where CONFIG_HEAP_TLSF_USE_ROM_IMPL defaults to y on
targets with ROM TLSF support like esp32c2), the TLSF patch sources
were being compiled into the bootloader's esp_rom. This caused a
build failure because esp_rom_tlsf.c includes tlsf_block_functions.h
from the heap component, which is not a dependency of esp_rom and is
not part of the bootloader build.
Guard the TLSF and multi_heap ROM patch sources with NOT
BOOTLOADER_BUILD to prevent them from being compiled in the
bootloader context. This fix is compatible with both cmakev1 and
cmakev2.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Commit f62f45cf5c changed target_add_binary_data to resolve the
embedded file path relative to the component directory by using
idf_component_get_property to obtain the COMPONENT_DIR. This fixed the
path resolution when target_add_binary_data is called from
idf_component_include, which runs outside the component's directory
context.
However, target_add_binary_data can also be called directly from a
project's CMakeLists.txt with a non-component target, e.g.
target_add_binary_data(app.elf ...) as done in
examples/security/security_features_app. In this case,
idf_component_get_property fails because the target is not a component.
Fix this by moving the path resolution to idf_component_include, where
the embed file paths from EMBED_FILES and EMBED_TXTFILES component
properties are resolved to absolute paths relative to COMPONENT_DIR
before being passed to target_add_binary_data. This way,
target_add_binary_data receives already absolute paths from
idf_component_include and can use plain get_filename_component(ABSOLUTE)
for direct calls, which correctly resolves relative to
CMAKE_CURRENT_SOURCE_DIR.
Fixes: f62f45cf5c ("fix(cmakev2/utilities): add a dependency target for the embedded file")
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add CMakeLists_v2.txt to the bootloader subproject, implementing the
bootloader build using the new cmakev2 IDF build framework.
The file covers the full bootloader build pipeline:
- Sets PROJECT_COMPONENTS_SOURCE to "idf_components" so that the
subproject's built-in components (main/, components/) are treated as
IDF components (priority 0) rather than project components (priority
3). This preserves the cmakev1 behaviour where user-supplied
components in bootloader_components/ can override the built-in ones.
- Registers optional user-supplied bootloader components from the
application project's bootloader_components/ directory, with support
for selectively excluding individual components via
IGNORE_EXTRA_COMPONENT.
- Bootstraps the cmakev2 framework (idf.cmake) and initialises the
project with BOOTLOADER_BUILD and NON_OS_BUILD properties, which are
also exposed as C preprocessor definitions.
- Sets GENERATE_SDKCONFIG to 0 to prevent the bootloader subproject
from regenerating the main project's sdkconfig, as the bootloader
has a different set of components and hence different Kconfig files.
- Sets the common implicit component dependencies shared by every
bootloader component (log, esp_rom, esp_common, esp_hw_support,
esp_libc, arch-specific component).
- Applies the compiler options specific for bootloader
- Selects the correct target-specific linker script, including a
separate script for ESP32-P4 silicon revisions < v3.
- Links the bootloader ELF via idf_build_executable and then converts it
to a flat binary via one of three paths depending on the secure boot
configuration:
* No secure boot: plain binary + size check + metadata.
* Secure Boot V1 one-time-flash: plain binary with post-build
instructions showing the esptool.py flash command.
* Secure Boot V1 reflashable: derives the symmetric eFuse key from
the ECDSA signing key, produces the reflash-digest image, and
prints burn/flash instructions.
* Secure Boot V2: produces an unsigned binary, optionally signs it
with the configured signing key (RSA-PSS 3072, ECDSA P-256, or
ECDSA P-384) via idf_sign_binary, and prints flash/multi-key
signing instructions.
- Adds comprehensive inline documentation explaining each section's
purpose, the rationale behind individual flags, and the relationships
between Kconfig symbols and generated artefacts.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add a GENERATE_SDKCONFIG build property that controls whether kconfgen
writes the sdkconfig file (--output config) during the configuration
step. It defaults to 1 (enabled).
When building the bootloader as a subproject, the set of components
(and their Kconfig files) differs from the main project. Running
kconfgen with --output config in this context rewrites the main
project's sdkconfig. Even when the content is identical, the timestamp
update causes ninja to detect sdkconfig as newer than build.ninja
outputs (e.g. cmake_install.cmake), triggering an infinite CMake
re-run loop.
Setting GENERATE_SDKCONFIG to 0 in a subproject prevents this,
matching the behaviour of cmakev1's __OUTPUT_SDKCONFIG property.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add the PROJECT_COMPONENTS_SOURCE build property to control how project
components are categorised during component discovery. By default it is
set to "project_components" (priority 3 - highest), preserving the
existing behaviour.
Setting it to "idf_components" (priority 0) before calling
idf_project_init() makes the project's built-in components overridable
by user-supplied components through EXTRA_COMPONENT_DIRS (priority 2).
This is needed for sub-projects like the bootloader, whose built-in
components (e.g. main) are provided by ESP-IDF and should be overridable
by user-supplied components placed in bootloader_components/. The
cmakev1 build system allowed this override because it used a
last-one-wins strategy, but cmakev2 uses strict priority-based component
resolution where project_components always win.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add an optional ALL parameter to idf_build_binary and idf_sign_binary
functions. When specified, the created custom target is included in the
default build target. Without ALL, custom targets created by these
functions are excluded from the default build (add_custom_target
behavior), meaning they won't be built unless explicitly requested or
depended upon by another target in ALL.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add a helper function for printing multi-line post-build messages on a
target. The message lines are joined with newlines, written to a file at
configure time, and printed at build time using a single cmake -E cat
command, replacing the need for repeated cmake -E echo invocations.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add a public API function to check that the bootloader binary does not
overlap the partition table, mirroring the existing idf_check_binary_size
pattern for application binaries.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The KEYFILE argument value was stored in a misspelled variable
"keyfle" instead of "keyfile", causing custom keyfile paths to
be silently ignored.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
`CMAKE_CURRENT_LIST_DIR` is actually `components/bootloader`, so it
doesn’t need to be passed via `EXTRA_COMPONENT_DIRS`: the build already
recognizes it as an esp-idf component.
In **cmakev1**, this is silently ignored: if a component with the same
name already exists, its directory is updated and the previous directory
is stored in the `COMPONENT_OVERRIDEN_DIR` component property.
In **cmakev2**, this is correctly detected and reported.
CMake Warning at /home/fhrbata/work/esp-idf/tools/cmakev2/utilities.cmake:63 (message):
IDF: Component 'bootloader' directory '/home/fhrbata/work/esp-idf/components/bootloader'
with higher priority 'project_extra_components' will be used instead of component directory
'/home/fhrbata/work/esp-idf/components/bootloader' with lower priority 'idf_components'
Call Stack (most recent call first):
/home/fhrbata/work/esp-idf/tools/cmakev2/component.cmake:625 (idf_warn)
/home/fhrbata/work/esp-idf/tools/cmakev2/idf.cmake:411 (__init_component)
/home/fhrbata/work/esp-idf/tools/cmakev2/project.cmake:580 (__init_components)
CMakeLists_v2.txt:28 (idf_project_init)
CMakeLists.txt:19 (include)
Since it doesn’t make sense to explicitly add the bootloader as an extra
component, remove it.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The commonly required components are configured solely for backward
compatibility with cmakev1. Since the bootloader build explicitly sets
commonly required components, we need to support this feature in cmakev2
as well. Ideally, there should be no commonly required components, and
each component should specify its requirements explicitly.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
For the bootloader build, the `priv_include_dirs` variable is
uninitialized, but it is used in the common call to
`idf_component_register`.
Using uninitialized variables is not allowed in cmakev2 because a
component may be evaluated in the context of another component’s
variable scope.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, the bootloader version of the `esp_partition` component sets
`PRIV_INCLUDE_DIRS` using the `private_include_dirs` variable. However,
this variable is not properly initialized, which causes issues in
cmakev2. In cmakev2, components are evaluated recursively, and a
component may be evaluated in the context of another component, so
components must initialize all variables before using them.
Moreover, there are effectively no `PRIV_INCLUDE_DIRS` to set when the
component is evaluated for the bootloader. Therefore, remove
`PRIV_INCLUDE_DIRS` entirely for bootloader and test builds.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Verify that setting IDF_COMPONENT_MANAGER=0 skips the component manager
flow entirely and produces a successful build. The test also asserts that
no "Component manager round" messages appear in the output, confirming
the manager loop is not entered.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
When the component manager is disabled via IDF_COMPONENT_MANAGER=0, the
cmakev2 build system still entered the full component manager flow
(__fetch_components_from_registry), which called
__download_managed_component(). That function returned early with
result=0 without creating the expected output file, violating the
post-condition in __download_component_level_managed_components() that
checks result==0 => file exists, causing a fatal error.
Instead of patching the low-level function to write an empty stub file,
this commit properly skips the entire component manager flow when the
manager is disabled:
- Move __init_component_manager() to idf.cmake global initialization
sequence alongside other __init_*() calls, so IDF_COMPONENT_MANAGER
build property is available early.
- Set __SDKCONFIG_ORIG to the real sdkconfig path in __init_kconfig()
as its default value. Previously it was only set inside
__create_sdkconfig_orig_copy() and had a defensive fallback in
__create_base_kconfgen_command(). The default ensures __SDKCONFIG_ORIG
is always valid regardless of whether the component manager runs.
- Guard __create_sdkconfig_orig_copy() with an IDF_COMPONENT_MANAGER
check. The sdkconfig backup exists solely to preserve unknown Kconfig
options from managed components during intermediate kconfgen rounds.
When the manager is disabled, no managed components exist, so the
backup is unnecessary.
- Guard __fetch_components_from_registry() call in project.cmake behind
IDF_COMPONENT_MANAGER == 1. When disabled, only the manifest warning
is issued. No download loop runs, no temp files are created, and no
"Component manager round N..." messages are printed.
- Remove the now-redundant IDF_COMPONENT_MANAGER guard from
__download_managed_component(), since it is only reachable when the
manager is enabled.
Closes https://github.com/espressif/esp-idf/issues/18372
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The cherry-pick in the backport of \!46287 (test/buildv2_pytest_build_v6.0)
dropped the deletion of tools/cmakev2/test/CMakeLists.txt. The original MR
deleted all 15 files from the tools/cmakev2/test/ directory, but the
cherry-pick only captured 14 deletions. This commit removes the missed file.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The ULP sources include sdkconfig.h for compile-time configuration
values. The sdkconfig directory was previously available to the ULP
target indirectly through the COMPONENT_INCLUDES variable, which is
populated from the parent component's INTERFACE_INCLUDE_DIRECTORIES.
In cmake v1, idf_component_register() adds the config_dir as PUBLIC
include directory to every component (component.cmake:498), so it
ends up in INTERFACE_INCLUDE_DIRECTORIES and gets passed to the ULP
subproject via COMPONENT_INCLUDES.
In cmakev2, the config_dir is added as a build-level property and
applied to components as PRIVATE (component.cmake:1053), so it no
longer appears in INTERFACE_INCLUDE_DIRECTORIES. As a result, the
sdkconfig directory is missing from the ULP target's include paths.
Add the sdkconfig directory explicitly to the ULP target's include
directories. Note that SDKCONFIG_HEADER is already passed to the ULP
subproject and its directory is already extracted into sdkconfig_dir
for linker script preprocessing (IDFULPProject.cmake:40).
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
In cmake v1, __build_process_project_includes() exports all build
properties as CMake variables before including project_include.cmake
files. cmakev2 was missing this step, causing components like ULP that
reference build properties as CMake variables (e.g. ${SDKCONFIG_HEADER})
to receive empty values.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The cmakev2 kconfig module sets sdkconfig output paths using internal
property names (__SDKCONFIG_HEADER, __SDKCONFIG_CMAKE, etc.), but
components like ULP read the public names (SDKCONFIG_HEADER,
SDKCONFIG_CMAKE). This results in empty values being passed to the ULP
sub-project, causing its CMake configure step to fail.
Add public aliases matching the cmake v1 property names for backward
compatibility.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, embedded files can be added in two ways. First, by calling
the `target_add_binary_data` function directly within a component.
Second, by passing `EMBED_FILES` or `EMBED_TXTFILES` to
`idf_component_register`, or in cmakev2 by setting the `EMBED_FILES` or
`EMBED_TXTFILES` component properties.
The source file for an embedded file is generated using
`add_custom_command`. When the embedded file is added directly in the
component's CMakeLists.txt file by using the `target_add_binary_data`
function, the `add_custom_command` command is evaluated in the same
directory context where the component target is created. As a result,
the generated embedded file can be used automatically as a file
dependency in component's sources.
However, when an embedded file is added via `idf_component_register` or
by setting the component property, the call to `add_custom_command`
inside `target_add_binary_data` occurs after the component has been
evaluated, specifically after the `add_subdirectory` call, within
`idf_component_include`. This means it is not created in the same
directory context as the component's target. In this case, the embedded
file dependency is not added to the component's target, and the embedded
file is not generated. This behavior is described in the
`add_custom_command` documentation [1].
> A target created in the same directory (CMakeLists.txt file)
> that specifies any output of the custom command as a source
> file is given a rule to generate the file using the command at
> build time.
To fix this issue, an explicit custom target for the generated embedded
file is created and added as a dependency of the component's target,
ensuring that the file is generated correctly.
[1] - https://cmake.org/cmake/help/latest/command/add_custom_command.html
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
In cmakev2, a component can be evaluated within the context of another
component, so it's important to properly initialize each variable used
by the component.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Pytest's default import mode (prepend/rootdir) imports test modules as
top-level modules keyed by filename. When two directories contain test
files with the same basename (e.g. test_sdkconfig.py in both
test_build_system/ and test_build_system/buildv2/), pytest tries to
register both as the module name "test_sdkconfig". The second collection
fails with "import file mismatch" because the module object already
cached in sys.modules points to the first file.
Adding __init__.py to the buildv2/ directory makes it a proper Python
package. Pytest then imports its test modules under the package
namespace (buildv2.test_sdkconfig), which is distinct from the
top-level test_sdkconfig, resolving the collision.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The sdkconfig file may contain configuration options defined in Kconfig
files of managed components. Since kconfgen runs before the component
manager fetches these components, the Kconfig definitions for managed
component options are not yet available. The kconfgen --output config
flag regenerates sdkconfig from kconfiglib's internal state, which only
knows about options with loaded Kconfig definitions. This causes unknown
options (i.e., those from managed components) to be silently dropped
from sdkconfig during intermediate regeneration rounds.
Note that kconfgen's --config flag (used for reading sdkconfig) only
performs deprecated option name replacement and does NOT drop unknown
options. The problem is exclusively in --output config, which writes a
fresh sdkconfig from the parsed Kconfig tree state.
Fix this by introducing a __SDKCONFIG_ORIG build property that provides
an indirection layer for the --config input path:
- Before the component manager runs: __SDKCONFIG_ORIG points to a copy
of the original sdkconfig (build/sdkconfig.orig), created by the new
__create_sdkconfig_orig_copy() function. This copy preserves all
original options, including those from managed components.
- During intermediate kconfgen runs: --config reads from the preserved
copy (so unknown options survive as input), while --output config
writes to the real sdkconfig (unknown options may be dropped there,
but this is harmless since kconfgen always reads from the copy).
- After the component manager completes: __SDKCONFIG_ORIG is reset to
point to the real sdkconfig and __BASE_KCONFGEN_CMD is rebuilt, so
that subsequent operations (menuconfig, save-defconfig, confserver)
read and write the actual sdkconfig file directly.
The flow is:
__create_sdkconfig_orig_copy()
-> __SDKCONFIG_ORIG = build/sdkconfig.orig
__generate_sdkconfig()
-> --config build/sdkconfig.orig --output config project/sdkconfig
__fetch_components_from_registry():
loop:
download_components()
__generate_sdkconfig()
-> --config build/sdkconfig.orig --output config project/sdkconfig
if success: break
endloop
-> __SDKCONFIG_ORIG = project/sdkconfig
-> rebuild __BASE_KCONFGEN_CMD
idf_create_menuconfig() / save-defconfig / confserver
-> uses --config project/sdkconfig (the real file)
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The idf_component_optional_requires compatibility function in cmakev2
currently attempts to mimic the behavior of cmakev1 by adding a
dependency only if a component is already included in the build. In
cmakev1, this is handled by checking the BUILD_COMPONENTS list, which is
created during an early evaluation phase. Because cmakev2 removed this
early phase to allow for configuration-based component dependencies, the
build system does not inherently know which components will be part of
the build beforehand. To compensate, the current compatibility function
relies on the TARGET_EXISTS generator expression to determine if a
component should be linked.
This approach poses a problem because generator expressions are not
evaluated until the generation phase, but cmakev2 processes linker
scripts during the configuration phase by recursively scanning targets
linked to a library. Because the scanner does not recognize and cannot
evaluate generator expressions, any component linked optionally through
generator expression is skipped. If that component carries a linker
script, the script is never added to the library interface, resulting in
build failures. Since cmakev2 aims to support multiple libraries, a
component might also exist globally, causing TARGET_EXISTS to evaluate
to true, yet still be missed during a specific library's recursive scan,
leading to the same omission of necessary linker scripts.
To resolve this, the implementation of idf_component_optional_requires
is changed to check if a component is known to the build system in
COMPONENTS_DISCOVERED build property. If so, it is explicitly included
and linked directly via its interface target. While this may pull more
components into a build than the previous generator expression method,
potentially increasing build times, it ensures that dependency trees are
fully visible to the library target scanner. This serves as a practical
middle ground that maintains compatibility with existing components and
ensures build stability. This approach allows for the gradual fixing of
idf_component_optional_requires usage in components for cmakev2, with
the eventual goal of removing its usage in cmakev2 entirely.
This change also fixes cases where idf_component_optional_requires is
used to conditionally add requirements based on configuration options.
In cmakev2, the presence of a configuration option does not guarantee[1]
that a component has been included via add_subdirectory, unlike the
behavior in cmakev1. With this change even constructions like
```cmake
if(CONFIG_VFS_SUPPORT_IO)
idf_component_optional_requires(PRIVATE vfs)
endif()
```
will now work in cmakev2, as the updated idf_component_optional_requires
explicitly includes the required component if it is available to the
build system.
Closes https://github.com/espressif/esp-idf/issues/18133
[1] https://docs.espressif.com/projects/esp-idf/en/latest/esp32/api-guides/build-system-v2.html#id9
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add a simple test to verify that the buildv2_test_app can be built for
the Linux target.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, cmakev2 is being tested only in backward-compatible mode by
using the existing cmakev1 tests with the cmakev2 test application. We
also need to add tests specific to cmakev2, and it is convenient to
reuse the existing build system testing framework. Let's add a `buildv2`
subdirectory to the existing `tools/test_build_system` directory and use
the `pytest_collection_modifyitems` hook to ignore tests in this
directory unless the `--buildv2` option is used.
Without the `--buildv2` option, only the existing cmakev1 tests are
executed and tests in `buildv2` directory are skipped. With the
`--buildv2` option, the existing cmakev1 tests run with the cmakev2
testing application for backward compatibility testing, and all cmakev2
tests within the `buildv2` subdirectory are also executed.
Note: we cannot use the `pytest_ignore_collect` hook, because the
`--buildv2` option is not known to the pytest, so the
`config.getoption('--buildv2', False)` returns always False. We would
likely need to add the `--buildv2` option in the conftest.py in the
esp-idf root directory.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
fix: lsadjf las jflasjfl aslfsald asl fsadlf sladsal jfsadfas
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
This adds a clear header to the pytest output, indicating which build
system version is currently being tested.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
In cmakev1, the initial order of components linked to the executable is
determined by the `__build_expand_requirements` function and stored as a
list in the `BUILD_COMPONENT_ALIASES` build property. Later, the
components from `BUILD_COMPONENT_ALIASES` are linked to the executable
using `target_link_libraries`. The issue is that
`BUILD_COMPONENT_ALIASES` contains components in a "reversed" order
compared to how they would typically appear on the link command line. In
other words, components (archives) deeper in the dependency chain are
placed first. As a result, cmake has to repeat the archives in the link
command more times than is actually necessary. This has the beneficial
side effect of placing libfreertos.a before libmain.a, allowing the
`app_main` symbol to be resolved.
In cmakev2, there is no predefined order because it does not perform
early evaluation. Instead, the entire link command is created based on
actual component dependencies using `target_link_libraries`.
Consequently, freertos must set the `app_main` symbols as undefined.
This is already done for non-Linux targets, but it must also be done for
the Linux target.
Another issue arises with macOS, which mangles C symbols by prepending
an underscore. We cannot use the same method as in
`components/linux/assert_func.c` for `__assert_func`, so for macOS,
`_app_main` is used. In the future, if more undefined symbols need to be
added for the Linux target, we should consider introducing a helper.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, idf_build_generate_metadata only accepts binary targets for
which it generates metadata (project_description.json). On Linux
targets, binary images are not generated, but we still need to generate
project_description.json. Extend the current function to accept both
executable and binary targets and ensure project_description.json is
generated when a Linux target is used.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Some components simply return when included in a build for the Linux
target. This is silently ignored in the `idf_component_include`, and the
component interface target is linked as requirement in
`idf_component_register`. This has the side
effect of such components being reported as included, with their
configuration displayed in the menuconfig for included components,
rather than in the submenu for excluded components. For example, the
`bt` component, if added as a dependency in `idf_component_register`,
will be displayed in menuconfig as an included component for Linux
target. The `idf_component_include` function sets the component's
`COMPONENT_REAL_TARGET` property to `NOTFOUND` in such situations. Use
this information when config.env is generated put such components into
excluded submenu. Note that we cannot avoid linking empty component interface
into library, because there might be recursive dependencies and at the
time when the component is included we might not now if it has a real
target or not.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, the CONFIG_APP_BUILD_GENERATE_BINARIES option is used to
determine whether binary images will be generated in
idf_project_default. At present, this option is enabled even for the
Linux target, where the esptool_py component is not included, preventing
the generation of binary images. To address this issue, verify if the
target for esptool_py is present.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
In cmakev2, the configuration for all components is included because the
complete configuration is generated very early, before any component is
actually evaluated. This poses a problem for the Linux target, which
does not have a controller Kconfig.in file, causing kconfgen to fail due
to the missing include. To address this, let's use the osource command,
which does not include the referenced Kconfig file if it does not exist.
Closes https://github.com/espressif/esp-idf/issues/18142
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The IDF_TOOLCHAIN build property is currently incorrectly set to the
default `gcc` value for the linux target, whereas it should be empty.
This misconfiguration causes confusion for components like `soc`, which
adjust toolchain options based on the
IDF_TOOLCHAIN(CONFIG_IDF_TOOLCHAIN_GCC) build property's setting.
When sdkconfig is generated, the IDF_TOOLCHAIN build property is passed
as an environmental variable to kconfgen, and the CONFIG_IDF_TOOLCHAIN
configuration option is set based on this variable. Additionally, the
CONFIG_IDF_TOOLCHAIN_GCC and CONFIG_IDF_TOOLCHAIN_CLANG configuration
options are set accordingly. Subsequently, CONFIG_IDF_TOOLCHAIN_GCC is
used in several places, such as `components/soc/project_include.cmake`,
to configure the toolchain (compiler flags) by invoking functions from
`tools/cmake/toolchain_flags.cmake`, which is included only for
non-linux targets. As a result the configuration fails, because
functions from `tools/cmake/toolchain_flags.cmake` are not available on
linux target.
Since the IDF_TOOLCHAIN cmake cache variable is actually set in the
`tools/cmake/toolchain.cmake` file, the IDF_TOOLCHAIN build property
should be set after the toolchain is initialized in cmakev2's project
initialization. Note that each toolchain file, except for linux,
includes `toolchain.cmake`, which in turn includes
`toolchain_flags.cmake`. This means the IDF_TOOLCHAIN cmake cache
variable is set for every target except linux, because the toolchain
file for linux is empty. As a result CONFIG_IDF_TOOLCHAIN is empty and
CONFIG_IDF_TOOLCHAIN_GCC not set as for cmakev1.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The `srcs` list variable is used without initialization when building
for linux target. Initialize it explicitly.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The `srcs` list variable is used without initialization when building
for linux target. Initialize it explicitly.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The dependency chain currently tracks component names that are included
recursively with `idf_component_include`. However, these component names
can be ambiguous because a component may be referenced by different
names, such as with a namespace. Additionally, `idf_component_include`
can accept anything that `__get_component_interface` accepts, meaning
even the component interface target can be used to include the component
in the build. To uniquely identify each component, use component
interface targets instead of names in the dependency chain.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
When version.txt does not exist and git describe fails (e.g. in
release archives or environments without git), IDF_VER was set to
the raw git_describe output which resolves to "-128-NOTFOUND",
causing esp_get_idf_version() to return a garbled string.
Add a fallback that constructs the version string from the
IDF_VERSION_MAJOR, IDF_VERSION_MINOR and IDF_VERSION_PATCH
variables when git describe is not available.
Closes https://github.com/espressif/esp-idf/issues/18240
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The minimal build property is simply a shorthand for `set(COMPONENTS
main)`. The issue is that there is currently no check to verify whether
the `main` component actually exists or is known to the build system.
If the `main` component is not present, print an error message along
with suggestions on how to fix this inconsistency.
Closes https://github.com/espressif/esp-idf/issues/18219
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
If `.gitlab/ci/rules\.yml` is changed, make sure the
`check-tools-files-patterns` pre-commit hooks is triggered to verify
that all `tools/*` files are covered by rules.yml.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The MINIMAL_BUILD build property and the COMPONENTS variable are both
used to determine the initial component list for the build.
Currently, if the COMPONENTS variable is set, the MINIMAL_BUILD logic
is ignored during component selection, but the MINIMAL_BUILD build
property remains set. This leads to an inconsistent state where
menuconfig displays information indicating MINIMAL_BUILD is active,
even though it was ignored in favor of the COMPONENTS variable.
Fix this by setting the MINIMAL_BUILD property to OFF if the
COMPONENTS variable is used.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
In some instances, the COMPONENTS_DISCOVERED build property is used to
walk through the component list and obtain component properties, such as
during configuration preparation or the inclusion of project_include
files. Since we know the component interfaces from the
COMPONENT_INTERFACES build property, we can switch to the faster
`__idf_component_get_property_unchecked` method to obtain component properties
in these cases. This change reduces the reconfiguration time by 0.4
seconds on my computer.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The commit f951ae5b18 ("feat(cmakev2): Added component validation
checks for ...") introduced support for validating component sources and
include directories. It heavily relies on iterating through the
registered components to obtain their properties. For instance, it
examines all components for each component source file to ensure that
the source files do not originate from other components. However, it
turns out that `idf_component_get_property` is a bottleneck in this
process, causing reconfiguration to take 3-4 times longer than before
the introduction of this check. Since we know of all the component
interfaces, we can bypass the checks performed by
`idf_component_get_property` and instead use the raw version,
`__idf_component_get_property_unchecked`, which operates much faster.
Following are hello_world example reconfigure times before and after.
before:
-- Configuring done (11.1s)
-- Generating done (0.2s)
after:
-- Configuring done (3.9s)
-- Generating done (0.2s)
A new library property LIBRARY_COMPONENT_INTERFACES_LINKED is added that
keeps component interfaces linked to the library.
Fixes: f951ae5b18 ("feat(cmakev2): Added component validation checks for ..")
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Retrieve the value of the specified component property quickly. The
existing `idf_component_get_property` function performs various checks
to identify the component interface target, which keeps the component
properties, and accepts a component name, target, target alias, or
interface. While `idf_component_get_property` uses a cache to identify
the component interface relatively quickly, it is still much slower
compared to the raw `get_property`. The
`__idf_component_get_property_unchecked` function provides a faster way how to
obtain component property if the component interface is already known.
It skips all the checks, so it must be used carefully.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Resolve ruff's UP031 errors related to the use of percent formatting for
strings and lines longer than 120 characters.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
As part of inter-procedural optimizations (IPA), the compiler may
perform tasks such as constant propagation for functions. This involves
generating a specialized version of a given function with a new symbol
name that includes a suffix. For example, during constant propagation,
the compiler might create a specialized version named
`spiflash_start_core.constprop.0` for the `spiflash_start_core`
function. Additionally, the compiler may generate multiple clones of a
single function. Currently, when ldgen performs symbol placement, it
does not account for these compiler-generated functions, leading to
their incorrect or unexpected placement in memory (markers).
Consider a linker fragment with:
```
[mapping:spi_flash]
archive: libspi_flash.a
entries:
esp_flash_api: spiflash_start_core (noflash)
```
The `spiflash_start_core` function should be placed in IRAM. However,
the compiler might generate an optimized version of this function with a
`.constprop.0` suffix, resulting in a
`.text.spiflash_start_core.constprop.0` input section. Currently, ldgen
does not handle this situation, leading to misplaced symbols.
Since `.` is not allowed in C identifiers, it should be safe to consider
all input sections for a symbol with any `.` suffix as representing that
symbol. This means considering the symbol suffixes should not cause any
ambiguity.
This change automatically places all input sections, including those
with possible suffixes for a given symbol, into the specified memory. In
other words, specifying a function name like `spiflash_start_core` in a
linker fragment automatically includes input section names matching
`spiflash_start_core(\..*)?$`.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The idf_build_generate_depgraph function creates a component dependency
graph in dot (graphviz) format for a specified executable. It uses
existing helper functions from cmakev1, ensuring that the generated dot
files are produced in the same manner as in cmakev1. While adjustments
might be needed in the future if necessary, the current implementation
is intended to offer the same functionality as cmakev1. Similar to
cmakev1, the dot files are only generated only when the
__BUILD_COMPONENT_DEPGRAPH_ENABLED build property is set.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The build system keeps track of each component source. Currently
there are four types of sources:
1. "project_components" - project components
2. "project_extra_components" - components from EXTRA_COMPONENT_DIRS
3. "project_managed_components" - custom project dependencies managed by the IDF Component Manager
4. "idf_components" - ESP-IDF built-in components, typically under /components
This can be used to identify the component libraries that are likely to
change during application development and pass them to ldgen as mutable
libraries. Add all components with "project_components" as their source
as mutable.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add a helper function `__should_generate_sdkconfig()` that checks
whether the sdkconfig file should be generated. The
`__generate_sdkconfig()` function can be called multiple times, such as
when the initial sdkconfig is generated at the start of the build
process and later after additional components are fetched by the
component manager. There might be no components fetched by the
component manager, for example, in the hello_world example, or the
downloaded components may not contain any configuration files. In such
cases, there is no need to regenerate the sdkconfig. This helper
function stores the list of configuration files in the
`__PREV_KCONFIGS`, `__PREV_KCONFIG_PROJBUILDS`, and
`__PREV_SDKCONFIG_RENAMES` build properties at its end, and at the
beginning, it compares them with the current lists of configuration
files.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The idf_create_size_report function allows for the creation of size
report targets based on the generated link map file. The size report
targets are created using the TARGET option name: "<target>",
"<target>-files", and "<target>-components". These size report targets
are added to the idf_default_project with the TARGET set to "size",
resulting in the creation of "size", "size-files", and "size-components"
targets for the default project.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The idf_project_init macro is evaluated within the global variable
scope, and currently, there are a few variables still set. Let's ensure
all variables are properly unset to avoid polluting the global variable
namespace.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The MINIMAL_BUILD property is not relevant in cmakev2, as only
components explicitly linked through targets are included in the build
by design. Display a warning when this outdated build property is set.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The link flag for generating the map file is currently added globally to
the link_options. If multiple executables are generated, the link map
file is overwritten by the last created executable. Since cmakev2
supports building multiple executables, the link map file options need
to be set for each executable individually. To address this, add a new
MAPFILE_TARGET option to the idf_build_executable function. When set,
the link map file will be generated by the linker, and a target
specified in the MAPFILE_TARGET option will be created for the map file.
This also splits the idf_project_default macro. Only the
idf_project_init macro needs be called within the global scope, as it
includes the project_include.cmake files and the cmake version of the
configuration. The remaining functionality of the idf_project_default
macro is implemented in a __project_default helper function to avoid
polluting the global variable space.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, the `idf_build_generate_metadata` function takes an
executable target as an argument, and the binary name is stored in the
`EXECUTABLE_BINARY` property. This approach is inconvenient because a
single executable might generate multiple binary images, making it
unreliable to store the binary image name in the executable property due
to the N:1 relationship. To address this, pass the binary image target
instead of the executable target to the `idf_build_generate_metadata`
function. This change is facilitated by linking the binary target,
executable target, and library interface targets with the following
properties:
binary:EXECUTABLE_TARGET -> executable:LIBRARY_INTERFACE -> library
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
With commit b719292b75 ("refactor(build-system): Simplify flash target creation"),
partition table and bootloader components are correctly adding their
dependency on the flash target, so there is no need for this to be done
in the build system. This removes the temporary workaround.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The `tools/test_build_system/buildv2_test_app` is a cmakev2 build system
testing application, same as `build_test_app`, which should be removed
from the readme check performed by check_build_test_rules.py.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The COMPONENTS and EXCLUDE_COMPONENTS variables are unused in cmakev2. A
deprecation warning will be printed if they are set. This change also
includes a helper functions for printing deprecation warnings.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The cmakev1 dropped generation of prefix_map_gdbinit, reflect this
change also in cmakev2.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently both menu configs, for component's project configuration and
component's configuration, for components not included in the build have
the same name: "Configuration for components not included in the build".
This might be confusing. Let's use "Project configuration for components
not included in the build" for component Kconfig.projbuild files.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The cmakev1 added esp_stdio to common requires, add it in cmakev2 too
for backward compatibility.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, the toolchain CMake files use the remove_duplicated_flags
function from utilities.cmake. The cmakev2 implementation also includes
this function for backward compatibility. Move the
remove_duplicated_flags function to a separate file,
deduplicate_flags.cmake, so it can be shared between cmakev1 and
cmakev2.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Enable check-kconfig-files. It was temporarily disabled because it was
not possible to add a new sourced file through an environment variable,
as kconfcheck had a fixed list of allowed environment variable names.
This issue should now be resolved.
Currently, the buildv2 tests are initiated using the same patterns as
the tests for the current build system. This means that any change in
the current build system will also trigger the buildv2 tests. Initially,
it might be wise not to block the CI for changes in the current build
system in case there is an issue with buildv2. Therefore, let's
explicitly start the buildv2 tests only when the buildv2 label is set.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The esp_libc relies on the stdio implementation of the vfs component if
CONFIG_VFS_SUPPORT_IO is enabled. This is not an issue in cmakev1
because if the vfs component is not included in the project build, its
configuration is not available, and CONFIG_VFS_SUPPORT_IO is not set.
However, in cmakev2, the configuration for all components is available,
and the presence of some component configuration options does not
necessarily mean that the component is included in the project build.
When esp_libc is compiled without the vfs dependency but with
CONFIG_VFS_SUPPORT_IO enabled, the libc initialization will encounter a
panic due to a NULL pointer dereference (fp->_flags) in __swsetup_r, as
fopen in esp_libc_init_global_stdio will return NULL.
Although the current documentation exposes only limited functionality,
which is unlikely to change, add an explicit note that the build system
v2 is currently a technical preview intended for testing and early
evaluation.
Ensure consistency in referring to v2 throughout the documentation. In
some places, it is referred to as V2 or Version 2; unify this to v2.
Additionally, update the hello_world example for v2 to use the
idf::spi_flash alias instead of the interface variable to maintain
consistency with the rest of the documentation.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Extend the current documentation to recommend using the
idf_component_register function as the preferred method for creating a
new component for v2.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Using idf_component_register is the preferred method for creating new
components for cmakev2. This approach ensures compatibility with both
versions of the build system. The KCONFIG and KCONFIG_PROJBUILD options
have been removed from the API documentation, but are retained in code
in case a cmakev1 component uses these options to warn about
incompatibility. Also remove a note about `project_include.cmake`,
because cmakev2 includes all project_include files.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Enhance the `esp_target_info` component example by adding some of
the missing features that the component might require. This includes
adding a linker script, a linker fragment, linking the component archive
as a whole library, and providing an example of how the linker options
can be set.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Components designed exclusively for cmakev2 are not compatible with
cmakev1 due to the lack of forward compatibility. It is important to
clarify this at the very beginning of the `Creating a New Component`
section. Additionally recommended a guidance on how to create a
component that will be compatible with both versions.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, the archive file name is set only in the cmakev1
idf_component_register shim. The predictable component archive file name
is important, for example, for usage in linker fragments. Ensure that
the cmakev2 component also has the archive file name set.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, when the cmakev2 component sources are set, the SRCS target
property is used. This is obviously wrong because the correct CMake
property is SOURCES.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Rename the LINKER_SCRIPTS_STATIC component property to LINKER_SCRIPTS.
This property stores linker scripts that are not processed by ldgen,
which essentially includes all of them. The only linker script processed
by ldgen is sections.ld, which is handled by the esp_system component.
This implies that there is likely no practical use case for other
components to utilize ldgen processed linker scripts. This change is
purely cosmetic to allow components to add linker scripts with:
idf_component_set_property(${COMPONENT_TARGET} LINKER_SCRIPTS linker_script.ld APPEND)
instead of
idf_component_set_property(${COMPONENT_TARGET} LINKER_SCRIPTS_STATIC linker_script.ld APPEND)
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Add very basic documentation. It is currently written with the
expectation that the reader is already familiar with the existing build
system and focuses more on the changes that might be necessary to adapt
existing components to work with cmakev2. Additionally, it provides
basic information on how to create a new component and project.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, we include numerous functions in the automatically generated
documentation for the build system API. Let's begin with only the
essential functions and gradually add more to the API based on requests
and actual needs.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The current approach involves wrapping the library target within the
library interface target with the whole-archive flags, such as for the
GCC linker. However, this does not work as expected because the library
target is expanded, and the whole-archive flags are also applied to
other library targets that the wrapped library target depends on. IOW
the whole-archive flags surround multiple archives, not just the one
requrested. Generally, using linker flags like whole-archive in the
library INTERFACE(INTERFACE_LINK_LIBRARIES) does not seem to work as
CMake may perform deduplication and rearrange the flags [1].
The proof of concept used `CMAKE_LINK_LIBRARY_USING_<FEATURE>` with the
WHOLE_ARCHIVE feature, which was introduced in CMake 3.24 to specify how
the library target should be linked. There are two issues with this.
First, this feature is only available from CMake 3.24 onwards, while our
minimum CMake version is set to 3.22. More importantly, all occurrences
of a library on the link line are wrapped. For example, if a library
like `vfs` appears multiple times on the link line due to dependencies,
each occurrence will be wrapped with whole-archive, causing the linker
to complain about multiple symbol definitions. Therefore, even though
WHOLE_ARCHIVE is recommended for handling whole-archive linkage, it does
not seem suitable for our purposes. This was overlooked in the PoC
because only a simple testing component was used to check the
WHOLE_ARCHIVE behavior.
One way to address this issue is to adopt the same approach used in
cmakev1, which involves specifying the whole-archive flags when linking
component library targets to the final executable. In this case, CMake
retains the flags without any alteration, unlike when the flags are
specified in INTERFACE_LINK_LIBRARIES for the component interface
targets. While this approach is feasible, it would alter the current
logic, where we have a single library interface for the entire idf
library that can be linked to the executable. This change would also
complicate the direct use of the idf library, as it would no longer be
possible to simply link it to the executable. Instead, the executable
would need to correctly link the component libraries with the
whole-archive flags. We could encapsulate this process within the
idf_build_executable function, but projects that only use
idf_build_library would need to implement the same solution.
It appears possible to address this issue by using target_link_options
instead. This allows to specify exactly what should appear on the link
command line. One side effect of this approach is that the library
appears on the link command line multiple times: first when used with
target_link_options and second when used in target_link_libraries.
However, this does not seem to pose a problem. The flags specified with
target_link_options appear on the command line first, ensuring that the
whole-archive is prioritized, and the component archive libraries may be
repeated on the link line anyway due to dependencies. Essentially, the
final link command line is very similar to the one from cmakev1. We may
revisit this approach in the future, but for now, it seems to work as
expected.
[1] https://gitlab.kitware.com/cmake/cmake/-/issues/20078
[2] https://cmake.org/cmake/help/latest/variable/CMAKE_LINK_LIBRARY_USING_FEATURE.html
[3] https://discourse.cmake.org/t/automatically-wrapping-a-static-library-
in-whole-archive-no-whole-archive-when-used-during-linking/5883
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, when the cmakev1 properties for INCLUDE_DIRS and
PRIV_INCLUDE_DIRS are constructed in the
__set_component_cmakev1_properties function, the return values from the
get_target_property function for INCLUDE_DIRECTORIES and
INTERFACE_INCLUDE_DIRECTORIES properties are not checked. If a component
target does not set e.g. INCLUDE_DIRECTORIES property,
get_target_property will return a value such as `include_dirs-NOTFOUND`.
This value is subsequently passed to __get_relative_paths in the PATHS
argument, causing the file(RELATIVE_PATH) call in __get_relative_paths
to fail with an error.
```
CMake Error at /home/fhrbata/work/esp-idf/tools/cmakev2/utilities.cmake:235 (file):
file RELATIVE_PATH must be passed a full path to the file:
include_dirs-NOTFOUND
```
Fix this by explicitly set the PATHS to an empty list before passing it
to the __get_relative_paths function if the property is not set.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The commit adb2d5deee ("feat(cmake): Produce warnings when component dependen..")
introduced additional checks for source files and include directories
used by a component that are located outside the component's directory.
If these files and directories belong to another component, a warning is
issued. This feature has not yet been implemented in cmakev2, so related
tests are temporary disabled.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
In cmakev1, certain arguments provided to the idf_component_register
function are stored as component properties. These properties are used
to generate the project_description.json file, which other tools rely
on. Since the idf_component_register function is obsolete in cmakev2, we
need to recreate component properties such as INCLUDE_DIRS,
PRIV_INCLUDE_DIRS, REQUIRES, and PRIV_REQUIRES, which were previously
provided by the idf_component_register function in cmakev1. To achieve
this, let's examine the component's real target cmake properties and
reconstruct them to maintain compatibility with cmakev1.
The real target cmake properties may include generator expressions,
which are ignored. This should be acceptable, as they were not addressed
in cmakev1 either, and handling them is likely not feasible.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Return the difference between two lists, meaning the elements that are
present in the first list but not in the second.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
This helper function removes entries containing generator expressions
from the given list. The list is modified in place.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
This is a helper function that returns a list of paths relative to a
given base directory for a list of input paths.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
When a linker script file with a .in extension is added using the
target_linker_script function, it is processed with the C preprocessor.
The linker scripts are preprocessed only once, even if they are used in
multiple libraries, because they are the same.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
During the initialization of a component in the __init_component
function, add the COMPONENT_BUILD_DIR property. This can be used to
store component-specific generated files, such as preprocessed linker
scripts.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The `includes` list variable is used without initialization. Initialize
it explicitly.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The component has been renamed, update the name in the common
requirements for the cmakev1 components.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The component can be referenced by multiple identifiers such as the
component name, which is derived from the component directory name,
component target, interface, or aliases. All component properties are
attached to the component interface target, which is also used for
declaring component dependencies. The cmakev2 build system has a
function called __get_component_interface, which is responsible for
identifying the component interface based on the given component
identifier. Since this function is called frequently, it needs to be
reasonably fast.
This introduces a new __idf_component_interface_cache INTERFACE target,
which serves as a mapping cache between component identifiers and the
component interface. The cache for each component is initialized in the
__init_component function, which introduces a component to the build
system. Currently, the component interface search is conducted by
examining the COMPONENTS_DISCOVERED and COMPONENT_INTERFACES lists
stored as build properties. Since the build system is aware of most
component identifiers, such as component name, target, and alias, during
component initialization, it can add mappings between component
identifiers and the component interface to a cache, which is built as
the components are initialized. This cache is used in the
__get_component_interface function instead of looking into the
COMPONENTS_DISCOVERED or COMPONENT_INTERFACES lists. This significantly
speeds up the component interface search and also makes the code much
simpler and more readable.
The component interface cache also completely replaces the existing
component name resolution, which was introduced because of the component
manager, and the cache used for resolved component names. This is
possible because all the necessary information is available during
component initialization when the component interface cache is
populated. The ambiguity of components is resolved based on component
source/priority.
Here is an example of the component interface mapping for the
espressif__led_strip component to the idf_espressif__led_strip interface
target. The component name, without the namespace, is referred to as the
short name. In this example, it is led_strip.
- espressif__led_strip -> idf_espressif__led_strip # name(directory name)
- idf_espressif__led_strip -> idf_espressif__led_strip # interface
- idf::espressif__led_strip -> idf_espressif__led_strip # alias
- _idf_espressif__led_strip -> idf_espressif__led_strip # real target
- led_strip -> idf_espressif__led_strip # short name
When another component with the same espressif__led_strip name is
initialized with a different priority, there is no need to change
anything in the cache, because the mapping stays the same.
The cache must be updated when two components share the same short name
but belong to different namespaces. This situation is likely uncommon.
For instance, consider espressif__led_strip and my__led_strip. If
my__led_strip has a higher priority, the cache is updated to reflect the
short name as follows:
- led_strip -> idf_my__led_strip
If both components have the same priority, the short name mapping for
led_strip is entirely removed. Conversely, if my__led_strip has a lower
priority, no short name mapping is added for it.
The short name is also added to the COMPONENT_SHORT_NAME property of the
component.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
This is a revised version of the existing __create_confserver_target
function. It creates a specified confserver target for a given
executable. The kconfig_menus.json file, used by the IDEs, is generated
when the confserver starts. This differs from the previous behavior,
where kconfig_menus.json was created globally along with other sdkconfig
formats. The reason for this change is that kconfig_menus.json contains
the Kconfig menu hierarchy and it is not just a flat option-value
format. It needs to accurately reflect which configurations for which
components are included or excluded. The kconfig_menus.json is generated
at `build/kconfig_menus.json`, where IDEs expect it. This means the file
is overwritten every time the server starts by kconfig_menus.json
version for given executable, so only one confserver can run at a time.
This is likely acceptable, as I don't believe it's possible to safely
run multiple instances of confserver due to the potential race
conditions when the sdkconfig file are generated.
In the future, we may include the location of kconfig_menus.json in
project_description.json so it can be easily identified by IDEs for
each executable.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Currently, when the initial sdkconfig is generated, the
kconfig_menus.json format is also created alongside other formats like
CMake, JSON, or C header. The kconfig_menus.json depends on the Kconfig
hierarchy and cannot be generated globally. It must be generated for
each executable to ensure that included and excluded components are
correctly positioned within the Kconfig menu hierarchy. Remove the
automatic generation of kconfig_menus.json.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The current base kconfgen command is stored in the __BASE_KCONFGEN_CMD
build property, and it includes the --env-file option pointing to the
`build/config.env` file. The `build/config.env` file does not separate
component configuration files into included and excluded components,
because it is created before any executable is added and we need it for
the initial sdkconfig generation. The global `build/config.env` can be
used for every sdkconfig format (cmake, json, txt), but not for the
kconfig_menus.json format, because kconfig_menus.json relies on the
Kconfig layout and hierarchy. Let's allow specifying the --env-file for
each kconfgen invocation and remove it from the base kconfgen command.
This allows you to specify an --env-file for each executable, each with
its own config.env file.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
The __run_kconfgen function does not take any arguments. The output
formats and related files are stored in the __KCONFGEN_OUTPUTS_CMD build
property.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
This is basically a renamed version of __create_save_defconfig_target.
The creation of the save-defconfig target is done in
`idf_project_default()`.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
For a specified executable, create a menuconfig target using the name
provided in the TARGET option. The function generates a config.env file
specific to the executable, where Kconfig files for components linked to
the executable and Kconfig files for components not linked to it are
separated. This separation allows for a clear visual distinction between
the configuration of components that are linked and those that are not
linked to the executable within the menuconfig.
Signed-off-by: Frantisek Hrbata <frantisek.hrbata@espressif.com>
Frantisek Hrbata
