1. Disable unnecessary TX/RX threshold interrupts to avoid potential trap in DMA transfer
2. Start TX/RX DMA transfer simultaneously to fit H/W spec and avoid potential RX FIFO overflow issue
In Mbed OS, page size is program unit, which is different than FMC definition.
After fixing page size, we can pass NVSTORE test (mbed-os-features-nvstore-tests-nvstore-functionality).
Originally, we use 2 H/W timers for us_ticker/lp_ticker, one for counting and the other for alarm.
With H/W timer running in continuous mode, we could use just one H/W timer for counting/alarm simultaneously.
The rework includes the following:
1. Remove ticker overflow handling because upper layer (mbed_ticker_api.c) has done with it.
This makes us_ticker/lp_ticker implementation more succinct and avoids potential error.
2. Refine timer register access with low-power clock source
1. With mbed OS 5, mbed_sdk_init() is ensured to call before C++ global object constructor.
2. Refine startup file with GCC_ARM toolchain related to this modification.
fire_interrupt function should be used for events in the past. As we have now
64bit timestamp, we can figure out what is in the past, and ask a target to invoke
an interrupt immediately. The previous attemps in the target HAL tickers were not ideal, as it can wrap around easily (16 or 32 bit counters). This new
functionality should solve this problem.
set_interrupt for tickers in HAL code should not handle anything but the next match interrupt. If it was in the past is handled by the upper layer.
It is possible that we are setting next event to the close future, so once it is set it is already in the past. Therefore we add a check after set interrupt to verify it is in future.
If it is not, we fire interrupt immediately. This results in
two events - first one immediate, correct one. The second one might be scheduled in far future (almost entire ticker range),
that should be discarded.
The specification for the fire_interrupts are:
- should set pending bit for the ticker interrupt (as soon as possible),
the event we are scheduling is already in the past, and we do not want to skip
any events
- no arguments are provided, neither return value, not needed
- ticker should be initialized prior calling this function (no need to check if it is already initialized)
All our targets provide this new functionality, removing old misleading if (timestamp is in the past) checks.
The use of mktime was causing a fault when called in interrupt handler because on GCC it lock the mutex protecting the environment, To overcome this issue, this patch add dedicated routine to convert a time_t into a tm and vice versa.
In the process mktime has been optimized and is now an order of magnitude faster than the routines present in the C library.
There is an easy default implementation of spi_master_block_write that
just calls spi_master_write in a loop, so the default implementation
of spi_master_block_write has been added to all targets.
Serial implementation uses different vector handlers for sync/async calls respectively. The issue can be reproduced with the following flow:
1. Register sync mode callback with Serial.attach().
2. Sync call with Serial.putc()/getc().
3. Change to async call with Serial.write()/read().
4. Change back to sync call with Serial.putc()/getc().
Now, vector handller is still for async mode, not for sync mode.
To fix it:
1. Introduce internal function serial_enable_interrupt() for both sync/async vector handler enable/disable.
Original HAL function serial_irq_set() is reduced to call it for sync mode vector handler enable/disable.
2. Introduce internal function serial_rollback_interrupt() to roll back sync mode vector handler at end of async transfer.
Cruz Monrreal
ccli8
Jimmy Brisson
Cruz Monrreal
cyliangtw
Przemyslaw Stekiel
Deepika
Martin Kojtal
Sam Grove
Vincent Coubard
Bartek Szatkowski
Christopher Haster
adustm