As a part of the v1.6 update, littlefs added proper handling for
cleaning up memory in the case of an error during mount. This took care
of a memory leak users were seeing. Ironically, it turns out the implementation
and user patterns in mbed-os was _relying_ on this memory leak to avoid a
double free in the same case of an error during mount.
The issue was that a failed mount would leave the LittleFileSystem class in a
state where it thought it was mounted, and later it would attempt to
unmount the filesystem. With the previous memory leak this would be
"ok", and the leaked memory would be freed. But with the fix in v1.6,
no memory is leaked, and the incorrect free triggers a hard fault.
Fixed to clean up state properly on failed mounts.
1. Added a flag to issue START command
2. Do not send START command inside i2c_start function as
the LPI2C hardware will issue a STOP on reception
of a NACK
3. Remove the i2c_address global variable, this is not
required
Signed-off-by: Mahesh Mahadevan <mahesh.mahadevan@nxp.com>
- Duty cycle can be now only disabled with duty-cycle-on flag if region supports duty-cycle.
If region does not support duty-cycle, this flag has no effect and duty cycle is always
disabled.
- Also introduced a new flag (duty-cycle-on-join) to disable duty cycle for JOIN requests.
This flag can be used for testing only and is used to speed up JOIN request testing as
backoff times for JOIN request are really long (easily several minutes per attempt).
This flag works in conjunction with main duty cycle setting. Disabling duty-cycle-on-join
works only if duty-cycle-on is disabled (or region settings have duty cycle disabled).
In #b0b0261 we changed the RX2 data rate to start from the highest data rate
available for the PHY rather than standard defined DR.
This introduced a regression, i.e., even when somebody changed the default RX2 data
rate to something usable for their environment, it didn't take any effect. As in
reset_mac_params() we override the data rate with max value possible for that PHY.
This commit restores the original behaviour and we always use standard defined
values.
Typo in the config file made the LwIP stack way too small on EFM32GG11 STK3701 target. Additionally, the EMAC thread stack was slightly too small when debug printing is on.
1. Remove the repeated_start flag and code as this is not needed
for the LPI2C module
2. Enable the SION bit on the I2C pins
3. Enable 22K Pullup option of the I2C pins
4. Update the 0 byte write implementation to ensure the START
command gets flushed out of the FIFO
Signed-off-by: Mahesh Mahadevan <mahesh.mahadevan@nxp.com>
To handle timer rollovers the test tests-mbed_hal-common_tickers_freq
calls intf->set_interrupt(0). For this to work correctly the ticker
implementation must fire an interrupt on every rollover event though
intf->set_interrupt(0) was called only once. Whether an interrupt will
fire only once or multiple times is undefined behavior which
cannot be relied upon.
To avoid this undefined behavior this patch continually schedules an
interrupt and performs overflow detection on every read. This also
removes the possibility of race conditions due to overflowCounter
incrementing at the wrong time.
Update the low power ticker wrapper code so it does not violate any
properties of the ticker specification. In specific this patch fixes
the following:
- Prevent spurious interrupts
- Fire interrupt only when the ticker times increments to or past the
value set by ticker_set_interrupt
- Disable interrupts when ticker_init is called
Christopher Haster
Jimmy Brisson
Mahesh Mahadevan
Cruz Monrreal
Donatien Garnier
Martin Kojtal
Amanda Butler
Juha Ylinen
Kimmo Vaisanen
Hasnain Virk
Oren Cohen
Timo Urhonen
Steven
Russ Butler