Add `TARGET_ARCH_<ARCH>` for each architecture (e.g. `TARGET_ARCH_CORTEX` for
Cortex M) to allow users to overwrite the target triple for a specific arch
from ~/.profile or ~/.bashrc (or the like) without overwriting it for all others
as well.
Since former ESP32 toolchain versions used POSIX threads, module `pthread` was required. The built-in `cxa_ctor_guards` had to be replaced since they used the `pthread_once` function for singleton objects initialization where the parameter `once` was of incompatible type with that provided by RIOT's `pthread` module. The current ESP32 toolchain version no longer uses POSIX threads. The dependency on module `pthread` as well as according C++ hacks can be removed.
In #12955 optimization was switched to O2 because with the '-Os'
option, the ESP32 hangs sporadically in 'tests/bench*' if
interrupts where disabled too early by benchmark tests.
Since it hasn't been reproduced since and in #13196 O2 was causing
un-explained hardfaults, since the aforementioned issue could not
be reproduced we switch back to Os by removing O2, as Os will be
used by default.
GPIO32 and GPIO33 are used during boot to start an 32.768 kHz XTAL if it is connected to these GPIOs. If the 32.768 kHz XTAL is not connected, these pins can be used digital IO. However, the 32.678 kHz XTAL has to be disabled explicitly in this case. Furthermore, the handling of GPIOs greater than GPIO31 had to be fixed in I2C software implementation.
When entering a sleep mode, all wake-up sources should first be disabled before the wake-up sources required for the sleep mode are then stepwise enabled again. Otherwise, an wake-up configuration of one sleep mode may affect the wake-up within another sleep mode.
The WiFi interface should be stopped before reboot or sleep. But stopping the WiFi interface disconnects an existing connection. Usually, esp_wifi_netdev tries to reconnect on an disconnect event. However, trying reconnect with a stopped WiFi interface may lead to a crash. Therefore, the stop event has to be handled.
Now, where the vendor files for light/deep sleep mode are added, function `pm_off` does not need to implement this mode by itself. Instead the existing deep sleep with disabled wakeup sources is used for pm_off.
`rtc_init` is used after light sleep to update the system time from RTC timer. The fix corrects a small difference of about 230 ms which would sum up with each wakeup from light sleep.
By using a custom EPOCH for the RTC implementation, we can extend the
range of the 32 bit counter based RTC by 118 years.
It also reduces the code size compared to the stdlib based POSIX functions.
To be able to define common configurations for all ESP CPUs, the CPU specific configuration cpu_conf.h has to include a common configuration. For that purpose cpu_conf.h in cpu/esp_common is renamed to cpu_conf_common.h and included in CPU specific configurations.
During the write access to the SPI flash, the IROM cache is not available and only code from the IRAM can be executed. Therefore, the code of file system implementations which access the SPI flash must reside in IRAM.
The same tool 'gen_esp32part.py' is used for the generation of partition tables on ESP8266 as well as n ESP32. The tool is therefore added to 'dist/tools/esptool'
With the new toolchain version required to fix issue #13133, the compilation of `examples/posix_socket` fails due to a C linkage error in `atomic_base.h`. The reason is that including `drivers/include/mtd.h` in `boards/esp32/board_common.h` inside the `extern C` block finally leads to including `atomic_base.h` inside the `extern C` block which in turn to the C linkage error for the template definitions in this file.
ESP32 can be clocked with either a 40 MHz or 26 MHz crystal. Since most boards use a 40 MHz crystal, the configuration was previously fixed to a 40 MHz crystal. This commit changes the crystal from 40 MHz to automatic detection, allowing boards with a 26 MHz crystal like the Sparkfun ESP32 Thing DEV to be used.
- Unecessary definitions are removed.
- Since the 48-bit RTC hardware timer uses a RC oscillator as clock, it is pretty inaccurate and leads to a RTC time deviation of up to 3 seconds per minute. Therefore, a calibration during the boot time determines a correction factor for the 48-bit RTC hardware timer. Function _rtc_time_to_us uses now this correction factor and converts a raw 48-bit RTC time to a corrected time in microseconds. Thus, the 48-bit RTC timer becomes much more accurate, but it can't still reach the accuracy of the PLL driven 64-bit system timer. The Advantage of using RTC over 64-bit sydtem timer is that it also continues in deep sleep mode and after software reset.
- If the 64-bit system timer is used to emulate the RTC timer, it uses the RTC hardware timer to continue its operation after software .
It is possible to use different timers as RTC timer for the periph_rtc module. Either the 48-bit RTC hardware timer is used directly or the PLL driven 64-bit system timer emulates a RTC timer. The latter one is much more accurate. Pseudomodule esp_rtc_timer controlls which timer is used. Only if esp_rtc_timer is enabled explicitly, the 48-bit RTC hardware timer is used. Otherwise the 64-bit sytstem timer is used to emulate the RTC timer.
The explicit call of rtc_init during the CPU start was removed because rtc_init is called within the function periph_init. The display of the system time at startup had to be placed after the call to periph_init.
There is an existing function that returns the system time in us as a 64 bit value. Converting this 64 value in us to a 32 bit value in ms is more easier and uses the complete 32 bit range. Using only the low part of the 64 bit system time in us and dividing it by 1e3 cuts the 32 bit range.
To control the log level and the format of the log output of SDK libraries, a bunch of library-specific printf functions are realized which map the log output from SDK libraries to RIOT's log macros.
The buffer[EHTHERNET_MAX_LEN] used in _esp_wifi_send to convert the iolist of the given packet to a plain buffer for the WiFi interface should not be on the stack to prevent the sending thread's stack from overflowing.
To avoid that murdock times out before tests/pkg_spiffs and tests/pkg_littlefs time out, the configured test timeouts for these tests is reduced to 200 seconds which should be enough. An ESP32 needs an average of 60 seconds for these tests, while an ESP8266 needs in average 100 seconds.
ESP32 nodes can crash during SPI Flash write operations if required parts of the code are not in the IRAM but in the cached SPI Flash memory, which is disabled during the SPI Flash write operations. Therefore, the code of the SPIFFS package and the VFS module are now stored in the IRAM.
To reduce the information that are printed at the console during the startup, special bootloaders are required that suppress the outputs which are only informational. The according bootloader has to be selected during the make process.
Startup information, including board configuration, is only printed when module esp_log_startup is used. This reduces the amount of information that is printed by default to the console during the startup. The user can enable module esp_log_startup to get the additional startup information.
The UART peripheral clock seems to be sporadically set to wrong value when the CPU clock is changed. In this case, the UART clock is not set to 115.200 kbps but to 96 kbps, so that the output in the console seems like garbage. This can also cause automatic tests to fail. Therefore, the CPU clock is only changed if CONFIG_ESP32_DEFAULT_CPU_FREQ_MHZ defines a different default CPU clock than the one already used at boot time.
If the user or the board definition doesn't enable `esp_wifi` or `esp_eth`, `esp_now` is defined as default netdev.
fixup! cpu/esp32: defines esp_now as default netdev
To be able to catch a core panic in debugger and to get the last output from asynchronous UART , e.g., if the stack is smashed, the system is not rebooted immediately anymore but breaks, which stops the execution in debugger or reboots the system after WDT timeout.
Log outputs generated by binary ESP32 SDK libraries are mapped to the ESP32's log module which supports colored and tagged log outpus. Tagged log outputs from SDK libraries are handled accordingly.
The implementation of `log_module` for ESP32 was changed from functions to a macro-based implementation to be able to use the bunch of macros for colored and tagget log output generation.
Renames crypto functions of ESP32 SDK in vendor code to resolve the conflicts between `wpa-supplicant` crypto functions and RIOT's `crypto` and `hashes` modules.
To resolve the conflicts between `wpa-supplicant` crypto functions (part of the ESP32 SDK) and RIOT's `crypto` and `hashes` modules, the crypto function headers from ESP32 SDK are added to vendor code and the crypto functions are renamed using the prefix `wpa_`.
A number of tests insist that the number of thread priority levels is 16. However, when using the WiFi interface, a number of high priority threads are required to handle the WiFi hardware. In this case, the number of thread priority levels must be 32. Solves the problem of tests `tests/shell`.
Removes the dependency of the module riot_freertos from module xtimer. This avoids that xtimer is used even if it is not really needed which in turn occupies the first timer device and tests/periph_timer fails.
ESP32 log output was always tagged with additional information by default. The tag consists the type of the log message, the system time in ms, and the module or function in which the log message is generated. By introducing module `esp_log_tagged`, these additional information are disabled by default and can be enabled by using module `esp_log_tagged`.
Log module of ESP32 supports colored log outputs when module `esp_log_color` is enabled. The generation of colored log outputs is realized by a extending the bunch of macros with an additional letter indicating the type of log message,
For the implementation of the colored log output, two versions of the bootloader are introduced, one version with colored log output and one version without colors.
If `SPI_CS_UNDEF` is given as the `cs` parameter, CS pin must not be handled by the driver. Furthermore, if `cont` parameter is true, CS pin must not be disabled at the end of one transfer.
The ADC2 controller of the ESP32 is used by the WiFi module. The GPIOs connected to the ADC2 controller cannot be used as ADC channels if the WiFi module is enabled. This is clarified by a note in the documentation.
Thin archives also cause a boot loop when using the flash module.
To prevent further surprises, disable thin archives unconditionally
until the cause for this behaviour is known.
For a yet unknown reason, both esp8266 and esp32 get stuck in a
reboot loop when thin archives are used.
As a workaround, disable thin archives for now if esp_wifi is used.
fixes#12258
* CPU files should already have 'CPU' defined by the board.
* Do not conditionally define CPU as it is not needed.
This is part of cleanup prior to moving the CPU/CPU_MODEL to
Makefile.features.
Instead of having a send buffer as member `esp_wifi` netdev, a local variable is used now as send buffer. This avoids the need for a locking mechanism and reduces the risk of deadlocks.
Receive call back function `_esp_wifi_rx_cb` is called from WiFi hardware driver with a pointer to a frame buffer that is allocated in the WiFi hardware driver. This frame buffer was freed immediately after copying its content to a single local receive buffer of the `esp_wifi` netdev. The local receive buffer remained occupied until the protocol stack had processed it. Further incoming packets were dropped. However, very often a number of subsequent WiFi frames are received at the same time before the first one is processed completely. Having the single local receive buffer to hold only one received frame, led to a number of lost packets, even at low network load. Therefore, a ringbuffer of rx_buf elements was introduced which doesn't store the frames directly but only references to the frame buffers allocated in WiFi hardware driver. Since there is enough memory to hold several frames, the frames buffers allocated in WiFi hardware driver aren't freed immediatly any longer but are kept until the frame is processed by the protocol stack. This results in a much less loss rate of packets.
Events of different type can be pending at the same time. Therefore it is not possible to use ascending identifiers for the presence of a pending event. Rather, each event type has to be represented by one bit. Thes bits ORed identify all types of pending events. In the esp_wifi_isr function all pending events are then handled in one call. Otherwise, some events might be lost.
Use the -gz option to compress ELF sections containing DWARF information.
This saves around 50% of disk space, without any side effects.
See https://gcc.gnu.org/onlinedocs/gcc-9.2.0/gcc/Debugging-Options.html#Debugging-Options
for more infomation on this option.
Some platforms have an outdated toolchain that does not support -gz so
the flag is blacklisted there. Even then, the results are quite impressive.
I used @cladmi's `buildtest` branch (https://github.com/cladmi/RIOT/tree/wip/du/buildtest)
with this change and compiled the `examples/default` application:
```
$ BUILD_IN_DOCKER=1 DOCKER="sudo docker" make -C examples/default buildtest-indocker
```
The size was obtained with:
```
$ find output -name "*.bin.bindirsize" -type f -exec tail -n1 '{}' \; | cut -f 1 | awk '{s+=$1} END {printf "%.0f", s}'
```
Results:
- Vanilla: 10328112 KB (~10GB).
- with -gz: 4982788 KB (~5GB).
This was inspired by #8496.
Module `newlib` is now used by default. Therefore, the separation of initialization of ctors and the newlibc is not needed any longer. Instead of calling `do_global_ctors` and `_init` separately, `__libc_init_array` is called. Explicit function `do_global_ctors` is removed.
While deleting multiple sectors in flash, interrupts were disabled over the whole time. Thus, deleting the entire flash led to the triggering of the watchdog timer and thus to a restart. Therefore, the interrupts and the cache are disabled only for the time of deleting a single sector. The same problem occurred for read and write large data sets.
Initializing the stdio file descriptors in global reent structure with newlib fake stdio file descriptors led to the problem that newlib stdio functions printf and puts were not working since they can't operate on these fake stdio file descriptors. Therefore, this initialization was removed. Now, the real stdio file descriptors as created automatically by newlib are used. Specific functions `printf`, `puts`, `getchar`and `putchar` are not required any longer and are removed now.
Modules newlib and newlib_syscalls_default are now used by default. Conditional compilations for MODULE_NEWLIB_SYSCALLS_DEFAULT as well as alternative code are removed completely.
printf and puts used ets_printf before. Unfortunately, ets_printf adds an additional \r for each \n which is not consistent with other RIOT platforms. As a result some automatic tests failed. Therefore, both functions write now character-wise directly to the UART interface.
Using a mutex for critical section handling with portENTER_CRITICAL and portEXIT_CRITICAL does not work for RIOT, as this function can also be called in the interrupt context. Therefore, the given mutex is not used. Instead, the basic default FreeRTOS mechanism for critical sections is used by simply disabling interrupts. Since context switches for the ESP32 are also based on interrupts, there is no possibility that another thread will enter the critical section once the interrupts are disabled.
The option value length of Ethernet addresses can be more than 6 byte in lwIP. Therefore, the max_len parameter is check to be greater than or equal to ETHERNET_ADDR_LEN.
Module esp_idf_heap is enabled in cpu/esp32/Makefile.dep depending on other modules. Since cpu/esp32/Makefile.dep is read after cpu/esp32/Makefile.include, the conditional definition of the linker options for the wrapper functions had to be moved from cpu/esp32/Makefile.include to cpu/esp32/Makefile.dep.
If module esp_idf_heap is used, the memory management functions _malloc_r, _realloc_r, _calloc_r and _free_r have to be overridden by wrapper functions to use the heap_* functions of module _esp_idf_heap. However, this can lead to multiple symbol errors for these functions for some applications. To solve this symbol conflict, _malloc_r, _realloc_r, _calloc_r and _free_r functions are renamed to __wrap_* and the linker options are extended by -Wl,-wrap option when module esp_idf_heap is used.
When standard C libraries are added to BASELIBS to group them together with all other modules, there are multiple definitions for the putchar function. The one that is defined writing to the UART as standard output and the one that is provided by the standard C libraries. To solve this symbol conflict, putchar and getchar functions that use the UART as standard output/input are renamed to __wrap_putchar and __wrap_getchar and the linker options are extended by -Wl,-wrap option.
When linking an application, symbol pthread_setcancelstate is not known in standard C libraries, even if the pthread module is linked. This is because the pthread module is grouped with all other modules, but not with the default C libraries when they are added to LINK_FLAGS. Therefore, standard C libraries have to be added also to BASELIBS to group them with all other modules.
Fixes the problem that the compilation of an applications can throw unknown symbol errors for functions that aren't use at all. Thus, it is possible to remove the warning for unknown symbols and the compilation can abort if there are real unknown symbols.
Although it isn't explicitly specified in API, gpio_read should return the last written output value for output ports. Since the handling of inputs and outputs is strictly separated by several registers in ESP32, gpio_read returned always the initial value of the input register. Therefore, a case distinction had to make. While for input ports the real value has to be read from the input register, the last written value for the output port has to be read from the output register.
Rational: the periph_common module is required by (most) other periph drivers
and also during startup of the CPU/MCU to run periph_init. The latter is only
required if other periph drivers are used, hence periph_common should be a
depency of periph_* modules and *not* of the CPU/MCU. This PR fixes that
by making periph_common a depency of periph_* and removing the explicit
include in the CPU/MCU implementation.
PREFLASHER/PREFFLAGS/FLASHDEPS are evaluated by the main Makefile.include.
Their value does not need to be exported.
Testing
-------
`git diff --word-diff` only reports `export` being removed.
`git show --stat` reports `16 insertions(+), 16 deletions(-)`
Which is the same amount as lines that where matching
`export[[:blank::]]\+VARIABLE` plus the newline that is said to have
changed.
FLASHER and FFLAGS are evaluated by the main Makefile.include or by file
included by it. Their value does not need to be exported.
This will also prevent evaluating 'PORT' for FFLAGS when not needed.
Testing
-------
`git diff --word-diff` only reports `export` being removed.
`git show --stat` reports `84 insertions(+), 84 deletions(-)`
Which is the same amount as lines that where matching
`export[[:blank::]]\+VARIABLE`.
UART devices are now configured using static array in header files instead of static variables in implementation to be able to define UART_NUMOF using the size of the array instead of a variable.
SPI devices are now configured using static array in header files instead of static variables in implementation to be able to define SPI_NUMOF using the size of the array instead of a variable.
I2C devices are now configured using static array in header files instead of static variables in implementation to be able to define I2C_NUMOF using the size of the array instead of a variable.
DAC pins are now configured using static arrays in header files instead of static variables in implementation to be able to define DAC_NUMOF using the size of these arrays instead of a variable.
ADC pins are now configured using static arrays in header files instead of static variables in implementation to be able to define ADC_NUMOF using the size of these arrays instead of a variable.
Functions that are used by ADC and DAC peripherals are moved to a new submodule periph_adc_ctrl. This is necessary to compile separate submodules for ADC and DAC.
`top_of_stack` isn't aligned down to the previous 16 byte aligned address. Furthermore, `top_of_stack` as well as `XT_CP_SIZE` are used unaligned in `cpu/esp_common/vendor/xtensa/portasm.S` in the address computation for the coprocessor save area, .
Aligning pointer `p` down to the previous 16 byte aligned address results in a wrong address of the coprocessor save area during the initialization of the thread context. This leads to wrong values and wrong positions of these values in the coprocessor save area in inital thread context.
Since ESP8266 doesn't have a coprocessor, this bug affects only ESP32.
During the flash step esptool.py gives the following warning:
WARNING: Flash size arguments in megabits like '16m' are deprecated.
Please use the equivalent size '2MB'.
Megabit arguments may be removed in a future release.
esptool.py v2.7-dev
This patch replaces '16m' with '2MB' to enable future compatibility.
Signed-off-by: Yegor Yefremov <yegorslists@googlemail.com>
Fix of #11354: Function '_write_r' of ESP32's newlibc does not write the output of function 'write(STDIO_FILENO, ...)' to the UART interface. To fix this problem, module 'newlib_syscalls_default' is now used by default. Function '_write_r' of module 'newlib_syscalls_default' uses 'stdio_write' which in turn uses 'uart_write' if module 'stdio_uart' is used which is now the default case for ESP32.
xSemaphoreTakeRecursive() returned before the fix: pdFALSE(equal to pdFAIL) when the call was successful in obtaining the semaphore
and pdTRUE(equal to pdPASS) when the call did not successfully obtain the semaphore.
According to freertos documentation:
"pdPASS Returned only if the call to xSemaphoreTakeRecursive() was successful in obtaining the semaphore"
"pdFAIL Returned if the call to xSemaphoreTakeRecursive() did not successfully obtain the semaphore."
Fixed it to return the correct value.
xSemaphoreTake() returned before the fix: pdFALSE(equal to pdFAIL) when the call was successful in obtaining the semaphore
and pdTRUE(equal to pdPASS) when the call did not successfully obtain the semaphore.
According to freertos documentation:
"pdPASS Returned only if the call to xSemaphoreTake() was successful in obtaining the semaphore"
"pdFAIL Returned if the call to xSemaphoreTake() did not successfully obtain the semaphore."
Fixed it to return the correct value.
The GPIO for RX has to be initialized as input before the GPIO for TX can be initialized as output. Otherwise it could lead to creash if RX GPIO was used as output before.
Function uart_set_baudrate which is only used internally was made static and renamed to _uart_set_baudrate to indicate that it is an internal function. Furthermore, an additional waiting for flushed TX FIFO added. The reconfiguration is now handled as critical section.
An additional _ for static symbols has been added by mistake and should be removed. This will make future merging with the reimplementation of ESP8266 easier.
The build system contains several instances of
INCLUDES += -I$(RIOTBASE)/sys/posix/include
This is bypassing the module management system, by directly accesing
headers without depending on a module. The module is the posix module.
That line is also added when one of the posix_* modules is requested.
According to the docs, the posix module provides headers only, but in
reality there is also inet.c.
This patch:
- Moves `inet.c` into `posix_inet`, leaving `posix` as a headers-only
module.
- Rename `posix` as `posix_headers` to make it clear the module only
includes headers.
- Makes `posix_*` modules depend on `posix_headers`, thus removing the
explicit `INCLUDES+=...` in `sys/Makefile.include`.
- Ocurrences of `INCLUDES+=...` are replaced by an explicit dependency
on `posix_headers`.
The default macros GPIO_PIN and GPIO_UNDEF do not have to be overridden. The GPIO_PIN macro definition was even wrong for 40 GPIOs without splitting into ports, even if that did not lead to erroneous behavior.
Fixes sporadic blocking of the wifi thread in esp_wifi_recv_cb function under heavy network load conditions when frames are coming in faster than they can be processed. Since esp_wifi_recv_cb function is not executed in interrupt context, the msg_send function used for ISR event can block when the message queue is full. With this change esp_wifi can be flooded with icmpv6 packets of maximum size without any problems over hours.
RX callback function should be register when WiFi has been connected to AP successfully and should be unregistered when WiFi disconnects from AP. Therefore, esp_wifi_internal_reg_rxcb is called now in event handler on event SYSTEM_EVENT_STA_CONNECTED. It is reset now on event SYSTEM_EVENT_STA_DISCONNECTED.
Before function wifi_connect is executed, starting the WiFi driver should have been finished. This is indicated by the WiFi driver by sending event SYSTEM_EVENT_STA_START. Function wifi_connect is moved therefore to the event handler for SYSTEM_EVENT_STA_START.
The size of received and transmitted frames was stored in an uint8_t, which did not allow to process frames larger than 255 octets. However, WiFi has an MTU of 1500 octets.
Adds a memset function `system_secure_memset` which cannot be optimized out by the compiler. It uses the libsodium approach of weak symbols. Function system_secure_memset calls the standard memset. Calling an empty function declared with weak attribute after the memset call, prevents the compiler to optimize it out. The overhead is only one function call.
Although ESP32 has four SPI controllers, only two of them can be effectively used (HSP and VSPI). The third one (FSPI) is used for external memory such as flash and PSRAM and can not be used for peripherals. FSPI is therefore removed from the API. In addition, the SPI0_DEV and SPI1_DEV configuration parameters are renamed SPI0_CTRL and SPI1_CTRL to better describe what they define and to avoid confusion with SPI_DEV (0) and SPI_DEV (1).
Although ESP32 has four SPI controllers, only two of them can be effectively used (HSP and VSPI). The third one (FSPI) is used for external memory such as flash and PSRAM and can not be used for peripherals. FSPI is therefore removed from the API. In addition, the SPI0_DEV and SPI1_DEV configuration parameters are renamed SPI0_CTRL and SPI1_CTRL to better describe what they define and to avoid confusion with SPI_DEV (0) and SPI_DEV (1).
cpu/esp32/include/periph_cpu.h overrides the default definition of adc_res_t from periph/adc with a definition which contains only four resolution, two new resolutions and two resolutions defined by the default definition of adc_res_t. This gives compilation errors if an application uses other resolutions. According to the documentation, adc_sample should return -1 if the resolution is not supported. All other CPUs override adc_res_t either to add new resolutions or to mark resolutions as unsupported. But they all allow to use them at the interface. Therefore, esp32 overrides now the definition of adc_res_t with all resolutions that are defined by the default definition of adc_res_t and new platform specific resolutions. It returns -1 if a resolution is used in adc_sample that is not supported.
This commit adds the ESP32 vendor libraries for WLAN to the BASELIBS variable. This avoids having to define an additional archive group in the LINKGFLAGS variable which contains these vendor libraries and again RIOT module archive files with the symbols that are refered by these vendor libraries.
Some ESP32 boards (like my SparkFun ESP32 Thing) have a main clock
crystal that runs at 26MHz, not 40MHz. RIOT appears to assume 40MHz.
The mismatch causes the UART to not sync properly, resulting in
garbage written to the terminal instead of log output.
I’ve added:
* A new board configuration constant ESP32_XTAL_FREQ that defaults
to 40, but can be overridden by a board def or at build time to
force a specific value (i.e. 26).
* Some code spliced into system_clk_init() to check this constant and
call rtc_clk_init() to set the correct frequency.
* A copy of the rtf_clk_init() function from the ESP-IDF sources.
Fixes#10272
The problem seemed to be a pipelining problem of write and read instructions when swapping the context. An isync instruction when returning from a context switch solves the potential pipelining problem.
Reason for the problem was that tast_exit function in thread_arch.c tried to release the thread a second time although it was already released in sched_task_exit. A simple check whether the thread is already released (sched_active_thread == NULL) solved the problem.
Xtensa newlib version requires pthread_setcancelstate as symbol. Therefore, the module pthread was always used, which in turn requires the module xtimer. The xtimer module, however, uses TIMER_DEV(0). Therefore, tests/timers failed for TIMER_DEV(0).
