ESP32 supports running a "deep sleep wake stub" when coming out of deep sleep. This function runs immediately as soon as the chip wakes up - before any normal initialisation, bootloader, or ESP-IDF code has run. After the wake stub runs, the SoC can go back to sleep or continue to start ESP-IDF normally.
Deep sleep wake stub code is loaded into "RTC Fast Memory" and any data which it uses must also be loaded into RTC memory. RTC memory regions hold their contents during deep sleep.
* As the SoC has freshly woken from sleep, most of the peripherals are in reset states. The SPI flash is unmapped.
* The wake stub code can only call functions implemented in ROM or loaded into RTC Fast Memory (see below.)
* The wake stub code can only access data loaded in RTC memory. All other RAM will be unintiailised and have random contents. The wake stub can use other RAM for temporary storage, but the contents will be overwritten when the SoC goes back to sleep or starts ESP-IDF.
* RTC memory must include any read-only data (.rodata) used by the stub.
* Data in RTC memory is initialised whenever the SoC restarts, except when waking from deep sleep. When waking from deep sleep, the values which were present before going to sleep are kept.
* Wake stub code is a part of the main esp-idf app. During normal running of esp-idf, functions can call the wake stub functions or access RTC memory. It is as if these were regular parts of the app.
The wake stub in esp-idf is called ``esp_wake_deep_sleep()``. This function runs whenever the SoC wakes from deep sleep. There is a default version of this function provided in esp-idf, but the default function is weak-linked so if your app contains a function named ``esp_wake_deep_sleep()`` then this will override the default.
If supplying a custom wake stub, the first thing it does should be to call ``esp_default_wake_deep_sleep()``.
It is not necessary to implement ``esp_wake_deep_sleep()`` in your app in order to use deep sleep. It is only necessary if you want to have special behaviour immediately on wake.
If you want to swap between different deep sleep stubs at runtime, it is also possible to do this by calling the ``esp_set_deep_sleep_wake_stub()`` function. This is not necessary if you only use the default ``esp_wake_deep_sleep()`` function.
All of these functions are declared in the ``esp_deepsleep.h`` header under components/esp32.
Wake stub code must be resident in RTC Fast Memory. This can be done in one of two ways.
The first way is to use the ``RTC_IRAM_ATTR`` attribute to place a function into RTC memory::
void RTC_IRAM_ATTR esp_wake_deep_sleep(void) {
esp_default_wake_deep_sleep();
// Add additional functionality here
}
The second way is to place the function into any source file whose name starts with ``rtc_wake_stub``. Files names ``rtc_wake_stub*`` have their contents automatically put into RTC memory by the linker.
The first way is simpler for very short and simple code, or for source files where you want to mix "normal" and "RTC" code. The second way is simpler when you want to write longer pieces of code for RTC memory.
Data used by stub code must be resident in RTC Slow Memory. This memory is also used by the ULP.
Specifying this data can be done in one of two ways:
The first way is to use the ``RTC_DATA_ATTR`` and ``RTC_RODATA_ATTR`` to specify any data (writeable or read-only, respectivley) which should be loaded into RTC slow memory::
