Inadequate locking in the esp_timer component allowed corruption
of the s_timers linked list:
1. timer_armed(timer) returns false
2. another task arms the timer and adds it to s_timers
3. the list is locked
4. the timer is inserted into s_timers again
The last step results in a loop in the s_timers list, which causes
an infinite loop when iterated. This change always locks the
list before checking if the timer is already armed avoiding
the data race.
Related to ESP32-C6 chip only because this chip can power down the digital domain
during the light sleep. And after wakes up, systimer gets resumed,
and the alarm value < count value, so it leads the alarm fired immediately.
We get one unnecessary interrupt at light sleep exit time.
Other chips do not power down the digital domain related to systimer.
All the partition handling API functions and data-types were moved from the 'spi_flash' component to the new one named 'esp_partition'. See Storage 5.x migration guide for more details
Timers, periodic or not, can now be restarted thanks to esp_timer_restart function.
This is done atomically, which can be used to feed a periodic timer, or simply change the period.
ESP32-C2 has a single group timer, thus it will use it for the interrupt watchdog,
which is more critical than the task watchdog. The latter is implement in
software thanks to the `esp_timer`component.
esp_light_sleep_start() will stall the other CPU via esp_ipc_isr_stall_other_cpu(). After stalling the other CPU,
will call esp_clk_... API which themselves take locks. If the other stalled CPU is holding those locks, this will
result in a deadlock.
This commit adds a workaround calling esp_clk_private_lock() to take the lock before stalling the other CPU.
When ESP32-C2 is paired with a 26 MHz XTAL, the systimer tick
frequency becomes equal to 26 / 2.5 = 10.4 MHz. Previously we always
assumed that systimer tick frequency is integer (and 1 MHz * power of
two, above that!).
This commit introduces a new LL macro, SYSTIMER_LL_TICKS_PER_US_DIV.
It should be set in such a way that:
1. SYSTIMER_LL_TICKS_PER_US / SYSTIMER_LL_TICKS_PER_US_DIV equals the
actual systimer tick frequency,
2. and SYSTIMER_LL_TICKS_PER_US is integer.
For ESP32-C2 this means that SYSTIMER_LL_TICKS_PER_US = 52 and
SYSTIMER_LL_TICKS_PER_US_DIV = 5.
This introduced two possible issues:
1. Overflow when multiplying systimer counter by 5
- Should not be an issue, since systimer counter is 52-bit, so
counter * 5 is no more than 55-bit.
2. The code needs to perform:
- divide by 5: when converting from microseconds to ticks
- divide by 52: when converting from ticks to microseconds
The latter potentially introduces a performance issue for the
esp_timer_get_time function.
After refactoring the target components (e.g. esp32) no longer contained any real functionality.
What remained in these components have been moved elsewhere and the component itself deleted from the
build system.
Some components were including esp_timer.h without declaring a
dependency on esp_timer component. This used to work due to a
transitive public dependency on esp_timer from freertos component.
Add explicit dependencies where needed.
Also some source files were using esp_timer functions without
including the header file. This used to work because esp_timer.h was
included from freertos port header file. This commit adds esp_timer.h
includes where needed.
1. Clean up clk usage in IDF, replace rtc_clk_xtal/apb_freq_get with
upper level API esp_clk_xtal/apb_freq
2. Fix small errors and wrong comments related to clock
3. Add clk_tree_defs.h to provide an unified clock id for each chip
Modify the NGed drivers to adopt new clock ids
Chen Yudong