uarte_enable function was not supporting a case when async
api was enabled but instance did not use it. Added runtime
check for async instance presence.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Driver was failing when interrupt mode was enabled for given
instance but interrupt driven TX part was not used. In that
case uart was not disabled after sending a byte which resulted
in continuous interrupt triggering. Added check for
fifo_fill_lock which is set when uart_fifo_fill is used.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
In case of the following sequence of UART events:
- UART_ENDRX
- UART_RXRDY
- TIMER_RXTIMEOUT
The application receives one more byte that was received,
due to RX counter alignment upon ENDRX event.
The proposed fix moves the RX byte counter alignment to the
RX timeout event handler.
Signed-off-by: Tomasz Chyrowicz <tomasz.chyrowicz@nordicsemi.no>
The context parameter used across device power management is
actually the power state. Just use it and avoid a lot of
unnecessary casts.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Avoid calling PPI driver when enhanced poll functionality is
disabled. Fixing a case when driver failed to compile when
enhanced poll is disabled for all instances.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
NRFX UARTE would write to user supplied buffer on IRQ without checking
whether or not the supplied buffer had available space left to write
one char
Signed-off-by: Arne Edholm <arne.edholm@assaabloy.com>
If there is a UARTE receive error (e.g. framing or break), the RXTO
event may never come. Check error event too, to avoid an infinite loop.
Signed-off-by: Jim Paris <jim@jim.sh>
The NRF UARTE has an undocumented feature that when you flush the RX
FIFO, the RXAMOUNT register is not cleared to zero if the FIFO is in
fact empty. This fix is correcting something that was most likely a
typo.
Signed-off-by: Petri Oksanen <petri@iote.ai>
Lowest power consumption can be achieved when uarte peripheral
is disabled when not used. In low power mode, need for both
directions is tracked and if both are no in use peripheral is
disabled. TX disabling is instant but RX requires flushing RX
fifo because data in hardware fifo is lost when peripheral is
re-enabled.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Refactored driver to prepare for low power extension. Functional
change is limited to handling of RX_DISABLED event which is now
generated from RXTO interrupt context after RX is stopped. Previously,
RX was not stopped after the transfer.
Rx flushing function contains hardware limitation workaround.
Workaround is applied only if flushed data is not discarded.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
If statement was unconditionally reading a field from async struct
while pointer to this struct may be null if asynchronous API is
enabled but given driver instance is not using it.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Gracefully shutdown the UARTE peripheral when the async API is used.
Failure to do so results in the driver being unusable when powered back
up as the required events (ENDTX & TXSTOPPED) are not set. This also
ensures that the last byte sent out via `poll_out` is properly output
on the serial line before powering down.
Fixes#31930.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Update the drivers power state knowledge immediately after calling
`nrf_uarte_enable`. This ensures that the state is correct regardless of
which path the function exits by.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Power consumption was still high after putting uarte device into off
state. It was caused by ENDRX interrupt that was triggered after
calling STOPRX. ENDRX event was called with 0 amount but interrupt
got triggered and fifo_read was starting RX again.
Added disabling RX interrupt before disabling UARTE and reenabling at
device activation.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Now that we generate a header that extern's all possible devicetree
based device struct we can remove DEVICE_DT_DECLARE and
DEVICE_DT_INST_DECLARE as they aren't needed anymore.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
- Remove SYS_ prefix
- shorten POWER_MANAGEMENT to just PM
- DEVICE_POWER_MANAGEMENT -> PM_DEVICE
and use PM_ as the prefix for all PM related Kconfigs
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
If interrupt handler contains while loop which depends on
uart_irq_tx_ready() and uart_fifo_fill is called inside this
loop then if TXSTOPPED event occurs while code executes the
loop then uart_irq_tx_ready() will return true but uart_fifo_fill
will fail to write any bytes. That is because fifo_fill_lock is
cleared in handling of TXSTOPPED. To solve that added clearing
the lock inside uart_irq_tx_ready() if STOPPED event is detected
there.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Use the devicetree node as the source of object name and other
information used when defining the device structure.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Added early return from uart_poll_out when uart device is not in
active state. Poll_out is used from many contexts by multiple users
and currently here is no mechanism in power management to disable
them (console, logger when going to inactive power state.
Additionally, moved setting new state when disactivating the device to
ensure that no poll_out will happen after disabling the UARTE.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Refactoring poll_out to be ready for handling preemption. uart_tx and
uart_fifo_fill modified so they are resilient to being preempted by
uart_poll_out.
Refactored uart_poll_out implementation to be common for interrupt
driven API and asynchronous API. In both APIs active state is detected
by evaluating state of ENDTX and TXSTOPPED events. If anyone is set it
means that new transfer can be started.
Patch is fixing existing issues:
- potential bytes dropping with flow control enabled
- busywaiting for ENDTX (asynchronous API) - poor performance
- potential bytes dropping during preemption
- potential uart_tx returning -EBUSY when interrupted poll_out
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Reset rx byte counters when hw timer/counter is disabled.
Fixes issue where disabling and re-enabling async uart rx can produce
"UART_RX_RDY" events with invalid data when power management is used.
Signed-off-by: Audun Korneliussen <audun.korneliussen@nordicsemi.no>
These are all the case that coccinelle cannot find as they are inside
macro declarations.
Fixed via:
git grep -rlz -E "\(struct device \*" |
xargs -0 sed -i 's/(struct device/(const struct device/g'
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Now that device_api attribute is unmodified at runtime, as well as all
the other attributes, it is possible to switch all device driver
instance to be constant.
A coccinelle rule is used for this:
@r_const_dev_1
disable optional_qualifier
@
@@
-struct device *
+const struct device *
@r_const_dev_2
disable optional_qualifier
@
@@
-struct device * const
+const struct device *
Fixes#27399
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
uart_rx_enable might be called after reception is stopped by lack
of buffers with flow control enabled. In that case there is a couple
of bytes kept in the HW FIFO. When HW TIMER is used for byte counting,
it is using RXDRDY event to count bytes. RXDRDY event is generated when
byte is received in RXD. RXD is the head of HW FIFO. When DMA ends and
ENDRX event is generated, transmitter still transmits until RTS line
goes high and transmitter reacts to it and stops transmitting. First
byte received after ENDRX lands in RXD (and generates RXDRDY event),
next by goes to RXD+1 (does not generate RXRDY event).
RXDRDY event after ENDRX is counted by the TIMER. If TIMER is reset
before restarting RX this byte is lost in the calculation. Because of
that byte counters cannot be reset before enabling RX thus move to
initialization. It applies only to HW counting but to keep it consistent
resetting was removed for both modes.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Added detection of uart_rx_buf_rsp call which happend to late when
uart is already disabled.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
When `CONFIG_UART_ASYNC_API` is enabled, cleanup any pending synchronous
receptions at initialization time. Pending receptions are present when
this driver is used with the default mcuboot application.
When `CONFIG_UART_ASYNC_API` is not enabled, RX is always enabled by
this driver. As mcuboot does not use the async API, and does not cleanup
the UART driver before jumping to the next application, that application
boots with RX still enabled. This results in a 500uA increase in current
consumption.
Fixes#26555
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Per the nrf9160 datasheet, UARTE must be disabled prior to configuring
PSEL.TXD, PSEL.RXD, PSEL.CTS, and/or PSEL.RTS. This also ensures that
any prior running code, such as a bootloader, which leaves the UART
enabled will not prevent the Zephyr application from changing the
configuration.
Make a similar change also to the uart_nrf_uart.c driver for the
nRF5 series.
Fixes 27080
Signed-off-by: Pete Skeggs <peter.skeggs@nordicsemi.no>
So far, when rx_disable was called then received data was discarded.
This is currently not according to the API but it is needed if
rx_disabled is called due to out of band information about end of
packet.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Cleaned up flow control configuration. Added support for using only
cts or only rts.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
This is a follow-up to commit cf7dd4981f.
When disabling RX, it is necessary to clear the RXSTARTED event after
the ENDRX_STARTRX shortcut is deactivated, as the event might already
have been generated at this point. If the event is not cleared and
the disabling of RX is done from the user handler called in the context
of the ENDRX interrupt, a spurious UART_RX_BUF_REQUEST event will be
generated (although RX is already disabled) for which a corresponding
call to uart_rx_buf_rsp() would fail, as the second buffer is already
set. Depending on the application implementation, this can result in
other unexpected problems.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
When the loop that tries to obtain the lock in poll_out() finishes
because of hitting the max number of trials (what normally should
never happen), force the lock to be taken instead of just giving up
with sending the data. The latter approach that was in use so far
could not deal with a situation when some thread was aborted while
keeping the lock. Other threads would then have no chance to send
anything with poll_out() until it was called from an ISR.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
Although it is not clearly specified in the UART API, this function
is supposed to return 0 when the TX interrupt is disabled, regardless
of whether anything could fit into the TX buffer or not (this can be
concluded from looking at implementations of several samples and tests
present in the tree, also almost all UART drivers are implemented this
way).
The problem in the uart_nrfx_uarte driver case is that the flag that
allows generation of the TX interrupt is not cleared directly in the
uart_irq_tx_disable() function but this clearing is deferred to the
end of the current transmission, in case one is in progress, and hence
it is done in the interrupt handler. Consequently, when the interrupt
handler is not entered between calls to uart_irq_tx_disable() and
uart_irq_tx_ready() (e.g. when both functions are called in a loop
executed in the interrupt context), the latter may return an invalid
value.
Fix this problem by additionally checking if the TX interrupt was
requested to be disabled, not only if it is already disabled in
hardware.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
In some cases (eg at high baud rate, no HW flow control, and when BLE
radio/ints running) data could be lost between when enough characters
have been RX'd to fill the DMA buffer and when the ENDRX event was
fired, where the the STARTRX task is invoked to start filling the next
buffer (which is set up earlier, but I think will not be filled until
STARTRX).
To fix this, the SHORT is enabled between ENDRX and STARTRX whenever the
'next' buffer is available, so that STARTRX is invoked automatically and
subsequent chars go into the next buffer via EasyDMA.
To make this work properly, uarte_nrfx_isr_async() now handles the ENDRX
event _before_ the STARTRX event.
There was also an issue in rx_timeout() where the received character
count (rx_total_byte_count) could be incremented greater than the actual
buffer size. This arises from rx_total_byte_count value coming from the
counting the RXDRDY events (either by PPI/timer counter or counting the
RXDRDY ints themselves) and so if chars are received in the rx_timeout()
(or before ENDRX is handled) the rx_timeout() could increment rx_offset
past the length of the buffer. This could result the remaining 'len'
being calculated incorrectly (an underflow due to unsigned - signed ,
where signed > unsigned).
To fix this, we now store the lengths of the buffers and don't invoke
the UART_RX_RDY callback when the buffers are full; its handled by
ENDRX.
(Also note that the buffer size should be available via the RXD.MAXCNT
register on the nrf, but this register is not exposed through the nrfx
HAL and is also double buffered, so it seemed clearer to just track the
buffer lengths explicitly here in the driver).
Signed-off-by: Marc Reilly <marc@cpdesign.com.au>
for fixup
Signed-off-by: Marc Reilly <marc@cpdesign.com.au>
Kernel timeouts have always been a 32 bit integer despite the
existence of generation macros, and existing code has been
inconsistent about using them. Upcoming commits are going to make the
timeout arguments opaque, so fix things up to be rigorously correct.
Changes include:
+ Adding a K_TIMEOUT_EQ() macro for code that needs to compare timeout
values for equality (e.g. with K_FOREVER or K_NO_WAIT).
+ Adding a k_msleep() synonym for k_sleep() which can continue to take
integral arguments as k_sleep() moves away to timeout arguments.
+ Pervasively using the K_MSEC(), K_SECONDS(), et. al. macros to
generate timeout arguments.
+ Removing the usage of K_NO_WAIT as the final argument to
K_THREAD_DEFINE(). This is just a count of milliseconds and we need
to use a zero.
This patch include no logic changes and should not affect generated
code at all.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>