Enabled heap_and_stack test for baremetal.
Added a test to check that global variables are initialised.
In mbed_sdk_boot:
- Added initialisation for mbed_stack_isr_start/size and mbed_heap_start/size for all toolchains.
- ARM toolchain:
- Added call to mbed_toolchain_init() to initialise global variables.
- Moved microlib initialisation code from mbed_retarget.cpp to mbed_sdk_boot.c.
- IAR toolchain: there is no equivalent to a software init hook that can be called. __low_level_init() was used but since this function is called before RAM initialisation, it cannot be used to initialize global variables. Defined a new __mbed_init() function called from IAR startup file instead.
As the ARM compiler is in GNU compatible mode, it defines __GNU__ which
(based on pre-processor condition in mbed_printf_implentation.h)
causes the printf functions to be wrapped using _wrap_* instead of
$Sub$$*.
This commit modifies the pre-processor
conditions to check for __GNUC__last in order to correctly
substitute the printf functions depending on the toolchain in use.
It also gets rid of $Super$$* substitution as it is not needed. $Super$$
is used to identify the original unpatched functions.
Missing substitutions for ARM compiler internal optimized "printfs" are
also added.
- By default, Mbed OS build tools use standard C library for all supported toolchains.
It is possible to use smaller C libraries by overriding the "target.default_lib" option
with "small". This option is only currently supported for the GCC_ARM toolchain.
This override config option is now extended in the build tool for ARM toolchain.
- Add configuration option to specify libraries supported for each toolchain per targets.
- Move __aeabi_assert function from rtos to retarget code so it’s available for bare metal.
- Use 2 memory region model for ARM toolchain scatter file for the following targets:
NUCLEO_F207ZG, STM32F411xE, STM32F429xI, NUCLEO_L073RZ, STM32F303xE
- Add a warning message in the build tools to deprecate uARM toolchain.
- NewLib-Nano C library is not supporting floating-point and printf with %hhd,%hhu,%hhX,%lld,%llu,%llX
format specifier so skipping those green tea test cases.
Added missing mbed_error_initialize function call in bare metal boot code which is used
to retrieve the crash report from RAM in case of any previous abnormal re-boot but this function call is present in RTOS boot code.
* Make mbed_error use bitwise MbedCRC call rather than local
implementation.
* Remove use of POLY_32BIT_REV_ANSI from LittleFS.
* Move some MbedCRC instances closer to use - construction cost is
trivial, and visibility aids compiler optimisation.
If the fault handler was hit before the stdio console was used and
initialised, the initialisation code caused a "mutex in ISR" trap,
stopping the register dump from happening.
Temporarily set the `error_in_progress` flag at the top of the fault
handler, and restore it before calling `mbed_error`. Take the
opportunity to suppress fault dumps on recursive crashes, much as is
done inside `mbed_error`.
Build successfully when `platform.stdio-convert-tty-newlines` or
`platform.stdio-convert-newlines` are set to `true` by ensuring
`isatty()` is also included when `platform.stdio-minimal-console-only`
is set to `true`.
* Make mbed_error use bitwise MbedCRC call rather than local
implementation.
* Remove use of POLY_32BIT_REV_ANSI from LittleFS.
* Move some MbedCRC instances closer to use - construction cost is
trivial, and visibility aids compiler optimisation.
The retarget code allocates an array of FileHandle* for console and file
handling (filehandles). A tiny target only needs a console (putc/getc).
There is no need for file handling.
The POSIX layer and the array of FileHandle* is not required for small
targets that only need a console ; this code is optionally compiled
out if the configuration parameter platform.stdio-minimal-console-only is
set to `"true"`.
The two character sequence %% is used in standard implementations of
printf to print a single %. This is because % is essentially printf's
escape character for format specifiers and as \% cannot work printf
uses %%.
Therefore to be compatible with string buffers containing
%%, minimal-printf also needs to only print a single %.
Ensure the file descriptor stdout is passed to `fputc` when the high
level C functions to print to the console are referenced.
This issue fixed only affected binaries built with the ARM toolchain.
Locating and checking the length of the `ARM_LIB_HEAP` region is an
extra task, when we're just interested in the base and limit. Looking at
only those saves 8 bytes of ROM.
More space could be saved if we ensured all targets had `ARM_LIB_HEAP`,
removing the need for this run-time fallback code.
When moving to the second heap region due to overflowing the first
region, the `_sbrk` implementation assumed the allocation would fit in
the second region, and didn't check for that overflowing too.
Problem revealed in `stats_heap` test with GCC 8 on K64F - the allocation
attempt for 1GiB crashed, as `_sbrk` indicated 1GiB was available at the
start of the second region.
second region.
Presumably older versions of newlib fault that allocation attempt before
passing to `_sbrk`.
While there, adjust the code to not use a separate static `bool`, saving
RAM. We can track with just one pointer, as order of the two regions is
fixed, and already relied on by newlib.
The ARMC6 compiler inadvertently introduces the `_scanf_mbtowc` symbol
to the build. The commit provides a weak definition of the symbol to
satisfy the linker whenever the symbol is included. This affects ARM
Compiler 6 version 6.12 and earlier.
The compiler error was previously observed when the MICROLIB library is
used with the uARM toolchain. However, the weak definition was put in
`mbed-os/rtos` which is not included when the bare metal profile is used.
Suggested-by: @kjbracey-arm
Replace the sleep_manager_can_deep_sleep() with !_deep_sleep_locked.
Indeed, if we know we've taken the lock because we're using us_ticker,
no need to do the early wake.
Updated comments accordingly.
When next SysTimer wake-up is scheduler far enough, always consider
that deep sleep may be entered and program an early wake-up.
So that even if deep sleep is only allowed some time later, it can be
entered. If not doing this, then the deep sleep would be prevented by
SysTimer itself and may not be entered at all.
This has been proved to happen in a simple blinly example.
* Modify Doxygen grouping of `drivers` Public/Internal APIs
* Correct classification of `mbed_events.h`
* Amend name of Doxygen group containing Device Key API
* Classify `CallChain.h` as public API and relocate file
* Remove Doxygen group from `equeue_platform.h` as it has no Doxygen compliant documentation
* Move USB target specific code back to `usb/device/targets`
* Change Doxygen groups structure, splitting first by Public/Internal
This commit also does the following:
* groups the documentation of related API
* moves `events/internal/equeue.h` to `events/equeue.h`
* merges `events/source/README.md` to `events/README.md`
Also includes:
* rename `mbed_sleep_manager.c` to `mbed_power_mgmt.c` to match its
header file
* create Doxygen groups for public and internal APIs
* use relative path to include header files where inconsistent
* update references to internal APIs throughout libraries
* update the copyright year for all modified files
* Fixed wrapper functions for IAR
* Fixed and renamed profile to minimal-printf.json
* Moved minimal-printf under platform
* Removed minimal-printf/mbed_lib.json
* Modified exporter template to work with partial profile
* Prevented optimization of printf to avoid compiler function substitution
Rajkumar Kanagaraj
Anna Bridge
Evelyne Donnaes
Martin Kojtal
Hugues Kamba
Antti Kauppila
Kevin Bracey
Przemyslaw Stekiel
Philipp Steiner
Marcin Tomczyk
Dominika Maziec
Laurent Meunier
George Psimenos
Andrew Chong
Kyle Kearney