This file is explicitly included by the espressif hal module. It's an
internal file provided by picolibc and newlib. Provide a stub to let
code designed for those to work with the minimal C library.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Picolibc's retargetable locking is based upon having the user own the lock
type (struct __lock, along with typedef struct __lock *_LOCK_T), and then
having the picolibc internal code only refer to this type via the _LOCK_T
pointer typedef, leaving the actual struct undeclared there.
Zephyr wants to use 'struct k_mutex' for this type; the initial picolibc
port handled this by trying to redefine the picolibc locking to use 'void
*' instead of 'struct __lock *' by including '#define _LOCK_T void
*'. Which 'works' as long as the Zephyr code doesn't actually include
picolibc's sys/lock.h.
A recent picolibc change to support POSIX stdio locking has picolibc's
stdio.h including sys/lock.h, which breaks Zephyr's hack.
To fix this, create a real 'struct __lock' type as
struct __lock { struct k_mutex m; };
Define all of the required picolibc locking API with this real type,
referring to the mutex inside without needing any casts.
This required switching the definition of the C library global lock from
K_MUTEX_DEFINE to the open-coded version, STRUCT_SECTION_ITERABLE_ALTERNATE
so that it has the correct type and still lands in the same elf section.
The only mildly inappropriate code left is that lock are allocated using
k_object_alloc(K_OBJ_MUTEX), which "works" because the size of 'struct
__lock` will exactly match the size of 'struct k_mutex' because of C's
struct allocation rules.
Signed-off-by: Keith Packard <keithp@keithp.com>
Switch to the common abort implementation so that we can use it
from the assert hooks to avoid undefined behavior.
Closes: 84824
Signed-off-by: Keith Packard <keithp@keithp.com>
Change the condition so that ESP32 SoCs are using `_heap_sentry` to
get actual run-time heap size.
Signed-off-by: Marek Matej <marek.matej@espressif.com>
GCC versions before 4.5 do not have the -fno-lto flag and give an error if
it is used. The flag may also be unavailable on various other non-GCC
compilers.
Previously, the picolibc CMakeList.txt script assumed that the flag was
available, and hard-coded it as an addition to the compiler flags.
This patch improves compatibility by making use of the already existing
"prohibit_lto" CMake target compiler property, which is set to the
appropriate flag to disable LTO support, or unset if the compiler lacks
such an option.
Signed-off-by: Joel Holdsworth <jholdsworth@nvidia.com>
Mostly a revert of commit b1def7145f ("arch: deprecate `_current`").
This commit was part of PR #80716 whose initial purpose was about providing
an architecture specific optimization for _current. The actual deprecation
was sneaked in later on without proper discussion.
The Zephyr core always used _current before and that was fine. It is quite
prevalent as well and the alternative is proving rather verbose.
Furthermore, as a concept, the "current thread" is not something that is
necessarily architecture specific. Therefore the primary abstraction
should not carry the arch_ prefix.
Hence this revert.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Previously, in libc-hooks.c, the signature of `_open()` was as
shown below.
```cpp
int _open(const char *name, int mode);
```
This conflicted with the signature of `_open()` from newlib,
which is
```cpp
int _open(const char *name, int flags, ...);
```
Moreover, the mode and flags field were reversed, but only for
the Xtensa architecture due to the `_open_r()` hook that is
present in `libc-hooks.c`.
This manifested itself via a call to `fopen(file, "w+")`, where
the expected flags should include `O_CREAT | O_TRUNC`, or
`0x200 | 0x400`. Instead, the unexpected flags passed to the
underlying `zvfs_open()` call were `0x1b6`.
This change corrects the function signature and order of the
arguments.
Signed-off-by: Chris Friedt <cfriedt@tenstorrent.com>
The compiler requires that much of the C library be built without using LTO
so that various symbols are available for use by generated code, including
things like memset and memcpy.
Add -fno-lto when building both picolibc itself as well as the Zephyr
interface bits.
Closes: #81674
Signed-off-by: Keith Packard <keithp@keithp.com>
Select TC_PROVIDES_POSIX_C_LANG_SUPPORT_R to keep Zephyr from
including the common libc implementation of the various _r APIs.
Signed-off-by: Fin Maaß <f.maass@vogl-electronic.com>
Increase the number of locks to match what the ARC MWDT libc requires
now. The library wants to have 2 locks per each available FILE entry,
and then some more. Also do not include an internal libc header, as
all that is needed from that header is a simple typedef.
Signed-off-by: Ilya Tagunov <Ilya.Tagunov@synopsys.com>
If CONFIG_POSIX_API is enabled, but CONFIG_POSIX_DEVICE_IO_ALIAS_WRITE
is not, the _write replacement should be provided to make stdio work.
Signed-off-by: Ilya Tagunov <Ilya.Tagunov@synopsys.com>
`_current` is now functionally equals to `arch_curr_thread()`, remove
its usage in-tree and deprecate it instead of removing it outright,
as it has been with us since forever.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
The clock_t and clockid_t types are already defined in ARC MWDT headers,
so they should not be defined again in Zephyr headers.
Signed-off-by: Ilya Tagunov <Ilya.Tagunov@synopsys.com>
Until POSIX layer decoupling is done, we still need the size_t type
definition across the codebase, but it should come through sys/types.h
and not through toolchain-specific headers.
Signed-off-by: Ilya Tagunov <Ilya.Tagunov@synopsys.com>
ARC MWDT library does not contain implementations for the reentrant
POSIX time functions, so common Zephyr implementations end up being
used instead. However, there are no declarations, so we just add them
here, unguarded, until a better place is found.
Signed-off-by: Ilya Tagunov <Ilya.Tagunov@synopsys.com>
The fileno macro defined in ARC MWDT headers conflicts with the fileno()
function defined in lib/posix/options/device_io.c. We should undefine it
and replace with a POSIX-compliant declaration and a weak definition.
Signed-off-by: Ilya Tagunov <Ilya.Tagunov@synopsys.com>
The minimal libc is currently built as a zephyr_library and will be
included in the final link line as such. However, the c_library property
will still be set as "-lc" (for most linkers) and will be added to the
link line. This effectively requires that a separate libc implementation
be available in the toolchain and makes it possible to accidentally pull
from the non-minimal libc.
This doesn't seem desirable, so try to prevent this by clearing the
c_library property if we are using the minimal libc.
Signed-off-by: Jonathon Penix <jpenix@quicinc.com>
Moving specs argument to compiler and linker properties so that the
compiler and linker in use can decide how the flags are mapped / handled
for the compiler and linker in use.
This avoids specifying `--specs=spec.picolibc` for clang which prints a
warning about an unused argument.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
Zephyr is a bare metal build where standard libs are disabled.
This means that c and runtime libraries must manually be linked in.
This has generally been handled by using CMake's link libraries handling
but the issue with that is both de-duplication but also library link
order.
Standard libraries must be linked at last location to ensure symbols
are always available, however this is not optimal with
target_link_libraries() because this would ultimately require every
library to know the c library to link with, which is not desired.
Therefore, setup standard C and runtime library linking in linker
CMake files for toolchains where this is required.
This commit expands the principle introduced with toolchain abstraction,
see PR#24851.
This means that a toolchain implementation may specify standard C,
runtime, C++, etc libraries, as well as their link order.
Because a property approach is used, then Zephyr modules, such as the
Picolibc module can adjust such properties.
An optional `zephyr_linker_finalize()` macro is called at the end of
Zephyr's CMakeList process and can be used by the toolchain
implementation to define the final linker invocation.
This aligns the linker handling flow to the principle introduced in
PR#24851 and improves the flexibility and robustness of Zephyr build
system.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
The previous code used casts of address of int to float pointer.
This is undefined behavior in C and may cause a compiler to
optimize code incorrectly.
I have replaced it with a union of int and float types,
eliminating the cast problem.
Signed-off-by: Lars-Ove Karlsson <lars-ove.karlsson@iar.com>
This file uses something that is defined in the toolchain files,
fix a build error by including that header
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
Select TC_PROVIDES_POSIX_C_LANG_SUPPORT_R to keep Zephyr from
including the common libc implementation of the various _r APIs.
Signed-off-by: Keith Packard <keithp@keithp.com>
POSIX_THREAD_SAFE_FUNCTIONS already selects POSIX_C_LANG_SUPPORT_R and
POSIX_C_LANG_SUPPORT_R alread selects all of the COMMON_LIBC_*_R values.
That makes the 'default y if POSIX_THREAD_SAFE_FUNCTIONS' clauses in these
symbols redundant.
Signed-off-by: Keith Packard <keithp@keithp.com>
ctime.c and localtime_r_utc.c may use POSIX-only functions; to ensure those
are visible from a POSIX-conforming C library, define _POSIX_C_SOURCE in
these source files.
Signed-off-by: Keith Packard <keithp@keithp.com>
NATIVE_APPlICATION (old native_posix) does not support bulding with the
any other library than the host libC (EXTERNAL_LIBC).
MINIMAL_LIBC_SUPPORTED defaulted to y always, and instead the depends
was set in the MINIMAL_LIBC choice.
This lead to the right library selected, but an incorrectly set
MINIMAL_LIBC_SUPPORTED.
Many tests filter based on MINIMAL_LIBC_SUPPORTED, so they were
not filtered out appropriately.
Let's place the depends in the right place to avoid this problem.
Also remove a redundant depends on for PICOLIBC and NEWLIB_LIBC
which already have the dependency in their respective _SUPPORTED
option.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
Add missing braces to comply with MISRA C:2012 Rule 15.6 and
also following Zephyr's style guideline.
Signed-off-by: Pisit Sawangvonganan <pisit@ndrsolution.com>
Commit 282f77e732f0a4bc859f559bc5a748f953a8140c removed the only place
defining LIBC_INCLUDE_DIR. Remove the corresponding use of
LIBC_INCLUDE_DIR from newlib/CMakeLists.txt as this setting is no longer
being defined anywhere.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
Since Zephyr itself does not currently support processes, but
conformant applications should still be able to link, add stubs
for the remaining POSIX functions in the POSIX_SIGNALS Option
Group.
The POSIX_SIGNALS Option Group is required for PSE51, PSE52,
PSE53, PSE54, and likely many other POSIX Subprofiles.
Signed-off-by: Chris Friedt <cfriedt@tenstorrent.com>
Implemented the following:
- `asctime_r()`
- `asctime()`
- `localtime()`
- `localtime_r()`
- `ctime()`
- `ctime_r()`
Specifically:
- the implementation of `localtime()` & `localtime_r()` simply
wraps around the gmtime() & gmtime_r() functions, the
results are always expressed as UTC.
- `ctime()` is equivalent to `asctime(localtime(clock))`, it
inherits the limitation of `localtime()` as well, which only
supports UTC results currently.
Added tests for these newly implemented functions.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
Add a config to use the custom _sbrk function, defined by a user.
It is possible that an application doesn't want to use the entire
remaining RAM for the heap.
Signed-off-by: Dawid Niedzwiecki <dawidn@google.com>
The `tests/posix/headers/portability.posix.headers.without_posix_api`
test fails to build in case of toolchain which uses minimal libc by
default (for example ARC MWDT toolchain).
It can be easily fixed by providing sighal.h handler in minimal libc
Signed-off-by: Evgeniy Paltsev <PaltsevEvgeniy@gmail.com>
Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
Implement the selection of the picolibc source in a Kconfig
choice as they are mutually exclusive.
The following table represents the possible choices, where 'X'
means that `PICOLIBC_SUPPORTED=n` and is guarded in the parent
level as `PICOLIBC=n`, so we do not need to care.
Module Toolchain C++ | Choice
0 0 0 | X
0 0 1 | X
0 1 0 | Toolchain only
0 1 1 | Toolchain only
1 0 0 | Module only
1 0 1 | X
1 1 0 | Toolchain/Module
1 1 1 | Toolchain only
The current implementation favors `PICOLIBC_USE_TOOLCHAIN` over
`PICOLIBC_USE_MODULE` whenever possible, that preference is
maintained in this implementation as well - the TOOLCHAIN
source will be the default choice when the toolchain supports
it or when the C++ is enabled. Otherwise, fallback to MODULE.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
Currently, `PICOLIBC_SUPPORTED` could be enabled even when
neither the toolchain has picolibc nor the picolibc module is
present, this can be the case when Zephyr is built with
external toolchain (non-Zephyr SDK) + external build system
where PICOLIBC module doesn't exist.
Fix this by having `PICOLIBC_SUPPORTED` depends on either
toolchain OR module, while taking the libc++ into
consideration.
Here's the full logic table of when the PICOLIBC should
be marked as supported.
libc++ TOOLCHAIN MODULE PICOLIBC_SUPPORTED
0 0 0 0
0 0 1 1
0 1 0 1
0 1 1 1
1 0 0 0
1 0 1 0
1 1 0 1
1 1 1 1
Co-authored-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
Since `PICOLIBC_USE_TOOLCHAIN` is available only if
`PICOLIBC` is enabled, it doesn't need to depend on the latter
again.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
This reverts commit b2243af32d.
PR #73978 introduced a regression.
Unfortunately this PR cannot be reverted without reverting also
Let's revert both PRs to stabilize main again towards the 3.7 release.
For more details on the issue see
https://github.com/zephyrproject-rtos/zephyr/issues/75205
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
Since Zephyr itself does not currently support processes, but
conformant applications should still be able to link, add stubs
for the remaining POSIX functions in the POSIX_SIGNALS Option
Group.
The POSIX_SIGNALS Option Group is required for PSE51, PSE52,
PSE53, PSE54, and likely many other POSIX Subprofiles.
Signed-off-by: Chris Friedt <cfriedt@tenstorrent.com>
In device init phase, it will call _mbedtls_init before malloc_prepare
as mbedtls has higher priority defined in SYS_INIT..
_mbedtls_init() will call psa_crypto_init() and malloc buffer,
but z_malloc_heap is not initialized, which will cause device hang.
Should call malloc_prepare() before _mbedtls_init to fix this issue,
so add new Kconfig to increase the priority of libc to deafult 30.
Signed-off-by: Maochen Wang <maochen.wang@nxp.com>
Remove two casts since the type was already the same.
Otherwise, the casts caused a warning with IAR tools.
Signed-off-by: Lars-Ove Karlsson <lars-ove.karlsson@iar.com>
The following heap-related Kconfigs shared by the picolibc &
minimal libc have been deprecated for more than 2 releases,
remove them and update the Kconfigs accordingly:
- `CONFIG_MINIMAL_LIBC_MALLOC_ARENA_SIZE`
- `CONFIG_MINIMAL_LIBC_REALLOCARRAY`
Cleanup the handling to support the deprecated malloc arena
size configuration values.
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
The following Kconfigs have been deprecated for more than 2
releases, remove them:
- `CONFIG_SUPPORT_MINIMAL_LIBC`
- `CONFIG_MINIMAL_LIBC_MALLOC`
- `CONFIG_MINIMAL_LIBC_CALLOC`
- `CONFIG_MINIMAL_LIBC_REALLOCARRAY`
Signed-off-by: Yong Cong Sin <ycsin@meta.com>
