Test was staring a TX transfer and aborting it after 300 us from
the timer handler. Test was assuming that ongoing transfer will
not finish in those 300 us. This might not be true for higher
baudrates. Instead of using fixed timeout, a value is calculated
from the baudrate and targets to abort the transfer after approx.
20 bytes of 95 byte long transfer.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Some time ago a new shim for nRF UARTE was added (uart_nrfx_uarte2.c)
which used nrfx_uarte.c driver underneath. It was supposed to support
nrf54x platforms. However, later on legacy driver (uart_nrfx_uarte.c)
was extended to support nrf54x platforms and it takes less code size,
has better performance and more features. Shim uart_nrfx_uarte2 will
no longer be supported. As new shim is the default and there is a
Kconfig to pick the legacy shim (CONFIG_UART_NRFX_UARTE_LEGACY_SHIM=y)
it cannot be deprecated in the normal way. Additional Kconfig option
is created (DEPRECATED_UART_NRFX_UARTE_LEGACY_SHIM) which is enabled
if CONFIG_UART_NRFX_UARTE_LEGACY_SHIM=n and it selects DEPRECATED.
A warning was also added to the CMakeLists.txt.
Patch removes use CONFIG_UART_NRFX_UARTE_LEGACY_SHIM in tests.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Add new target to the following tests:
- uart_elementary
- uart_async_api
- uart_mix_fifo_poll
- uart_pm
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Due to cm7 on RT1180 is secondary core, and dtcm have been used
as sram directly, so there is no need to define zephyr,dtcm node.
Signed-off-by: Lucien Zhao <lucien.zhao@nxp.com>
Enable device runtime PM for nrf54h20dk/nrf54h20/cpuapp where
fast UARTE (uart120) is tested.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
With the changes made in PR #76061, only on the F429ZI platform, in
the test_write_abort test,the number of received data (2) is not equal
to the number of sent data (1). It looks like uart_tx_abort is done very
quickly, and the callback doesn't have enough time to deal with it.
Adding a small delay solves the problem and doesn't affect results on
other platforms.
Signed-off-by: Fabrice DJIATSA <fabrice.djiatsa-ext@st.com>
Test cases on F746ZG work with USERSPACE config not enabled.
PR #80944 forces the use of the TEST_ENABLE_USERSPACE config with
the "select" statement instead of "imply," which is more restrictive
for test case dependencies.This leads to the error
"Memory region 0x2004d180 (size 8) write access denied" on boards.
By setting CONFIG_TEST_USERSPACE=n, the TEST_ENABLE_USERSPACE config will
not be enabled.
Signed-off-by: Fabrice DJIATSA <fabrice.djiatsa-ext@st.com>
Convert qspi and hyperflash to variants instead of revisions by popular
demand.
And convert evkb into a revision instead of a different board.
Signed-off-by: Declan Snyder <declan.snyder@nxp.com>
Deprecate mode which is using TIMER+(D)PPI for reliable byte counting.
Recently a new approach is added (CONFIG_UART_NRFX_UARTE_ENHANCED_RX)
which supports reliable byte counting without additional HW resource.
This mode is planned to be the only supported RX path mode.
Enhanced RX has slightly different behavior. There are no partial RX
packets (events with non-zero offset). There is UART_RX_BUF_RELEASED
after each UART_RX_RDY event.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Enable this test in the simulated nrf54l15 by providing
an appropriate overlay.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
Add second instance to be tested on nrf54h20dk. uart120 is a fast UARTE
which works on fixed pin locations. It is not available for cpuppr core.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Enable this test in the simulated nrf5340 by providing
an appropriate overlay.
Signed-off-by: Alberto Escolar Piedras <alberto.escolar.piedras@nordicsemi.no>
nrf54h20dk_nrf54h20_cpurad by default has less RAM dedicated for
DMA transfers and default 1k buffer cannot be used for
uart_async_long_buf case. Use custom value.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
One of the test case is performing long transfers using 1k buffers.
For some targets there may be not enough RAM to perform such
transfers. Make long buffer length configurable.
Signed-off-by: Krzysztof Chruściński <krzysztof.chruscinski@nordicsemi.no>
Updated overlay files to test using nocache.
and run outside of usersapce mode to avoid
inaccessible memory regions when using nocache on RT parts.
Also updated testcase file to reflect nocache testing.
Signed-off-by: Emilio Benavente <emilio.benavente@nxp.com>
