Fix issue where an invalid (all zeroes) hash was written to settings
storage on reset. This caused the old value to written to zeroes, before
being written back to it's original value again immediately after.
This causes excessive flash wear.
This happens because the check if (k_delayed_work_remaining) returns the
amount of time until the work will execute. When that time has run out
the time is zero, but the work has not yet been executed.
We then write the invalid hash to flash, and then once the work-item
executes it will write the correct value.
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
Document that all services that should be included in the initial
database hash should be registered before calling settings load. All
services added after settings_load will trigger a new hash calculation
and a new value stored. This would result in the database hash always
being rewritten.
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
- avoid spourious radio interrupts by fixing ISR set,
waiting for idle, command configuration
- adjust counter to account for missing increment
- change preemption instant to avoid missing the deadline in LLL
- decrese EVENT_JITTER_US and
EVENT_TICKER_RES_MARGIN_US (same as Nordic)
Continuous scanning and connections are working fine now.
Signed-off-by: George Stefan <george.stefan@nxp.com>
Created unit tests for the encryption and decryption functions.
Tested with the peripheral and with central_hr samples.
Due to latency of CAUv3 when used as CCM inline accelerator
only one of the PDU can be encrypted/decrypted within an
bilateral exchange M->S + S->M in a connection event.
If the RXed PDU is encrypted, the TXed PDU must be empty
with More Data if there is data in the LLL queue.
The TXed PDU will be encrypted when an empty PDU is RXed.
Signed-off-by: Cristi Caciuloiu <cristian.caciuloiu@nxp.com>
Disabling gpio interrupts on the pin prior to changing its interrupt
type to level-based helps prevent spurious interrupts that would be
otherwise observed, if gpio interrupts were originally enabled when
gpio_cc32xx_pin_interrupt_configure() is invoked.
Fixes#22847
Signed-off-by: Vincent Wan <vincent.wan@linaro.org>
In the gpio isr, gpio interrupts are disabled when invoking the
registered callbacks. This is unnecessary, and causes a problem if the
callback attempts to disable gpio interrupts by reconfiguring the pin
as in the test gpio_basic_api.
Fixes#22847
Signed-off-by: Vincent Wan <vincent.wan@linaro.org>
The Runtime API and the name processing API were omitted in
the documentation as were not included in any doxygen group.
This patch includes them in dedicated settings API doxygen
sub-groups.
Signed-off-by: Andrzej Puzdrowski <andrzej.puzdrowski@nordicsemi.no>
Include gpio_utils.h to fix a build failure:
gpio_pca95xx.c:490:20: warning: implicit declaration of
function 'GPIO_PORT_PIN_MASK_FROM_NGPIOS'
[-Wimplicit-function-declaration]
gpio_pca95xx.c:490:20: error: initializer element is not constant
.port_pin_mask = GPIO_PORT_PIN_MASK_FROM_NGPIOS(
DT_INST_##inst##_NXP_PCA95XX_NGPIOS), \
Signed-off-by: Göran Weinholt <goran.weinholt@endian.se>
This is aligned with the documentation which states that an error shall
be returned if the work has been completed:
'-EINVAL Work item is being processed or has completed its work.'
Though in order to be able to resubmit from the handler itself it needs
to be able to distinct when the work is already completed so instead of
-EINVAL it return -EALREADY when the work is considered to be completed.
Fixes#22803
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Right now when building a release candidate in master cmake reports the
version wrongly as the final version, for example:
-- Zephyr version: 2.2.0
This is misleading and confusing. Cmake does not like the rcX suffix and
internally we indeed use 2.2.0 as the version.
This patch just changes the output of the status message and adds the
extra version field:
-- Zephyr version: 2.2.0-rc1
and continues to use the cmake compatible version internally.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Those checks are now being run using GH actions. ci-tools was changes to
support excluding modules and stopping them from reporting pending
status.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Encode all types of EDTT command response with similar structure to how
it is generated from command complete when it is generated from the
command status.
This structure will be:
2 byte EDTT command response code
2 byte EDTT length field
1 byte HCI status
N bytes of command output parameters
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
Enable tests that was disabled due to test failure. Test have been
updated and are now passing.
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
Now that we found the permission issue that was causing the aws s3 sync
to fail we can remove the debug flag.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
This branch filter doesn't work as expected with tag. Need to test this
more on testing tree to find a combo or solution that works properly.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
For some reason the aws s3 sync fails, but not sure exactly why. Adding
more debug to see if we can understand what's going on.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Limit doc publish workflow to master branch. Need to see if we can
seperate out 'latest' to be on master only and release branches to
publish to the proper location.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Update Atmel SAM-BA link address to MHCP SAM-BA web page. Now user will
navigate to right page instead be at MHCP home page.
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
The doc publish workflow is failing in the AWS S3 push. Add more
debug output in that section to see what's going on.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Fix up the board definition so that the actual board can at least be
processed by Kconfig.
See #22474.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
In the default configuration we want to enable the generic GPIO module
so that we avoid Kconfig warnings.
See #22474.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
In order to overrdie a choice one needs to define it again. Override it
by redefining it in the .defconfig file.
See #22474.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
For the non-secure image, the architecture still supports secure
execution, so express that in the defconfig.
See #22474.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
We need to unlock IRQs in early return points of
z_fatal_error() functions; not only at the normal
return point.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
Fix regression due to addition of conditional compilations
while porting the privacy feature from legacy to split
controller.
Fixes#22801.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
This works around an issue with this emulator's configuration where
there is no memory address that can be poked to generate a fault,
it is simulating memory for the entire address space.
Fixes: #22561
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The newlib full malloc implementation (i.e. non-nano) requests a
relatively large 4096-byte memory chunk through `_sbrk`, which exceeds
the configured 512-byte heap size.
This commit changes `CONFIG_NEWLIB_LIBC_ALIGNED_HEAP_SIZE` from 512 to
8192 in order to increase the size of the heap memory used by the
newlib malloc function.
For more details, refer to the issue #21167.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
NEWLIB_LIBC_NANO defaults to y when building with a toolchain that
supports nano.specs and this was causing the libraries.libc.newlib
test to link with the newlib nano variant (libc_nano.a) when it should
be linking with the normal newlib (libc.a).
By setting CONFIG_NEWLIB_LIBC_NANO=n in prj_newlib.conf, we make sure
that the libraries.libc.newlib test links with the normal newlib.
For more details, refer to the issue #21167.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit fixes the improper naming of the ROM sections.
1. Rename the first ROM section, which was previously named using the
`_TEXT_SECTION_NAME` definition, to `rom_start`, as this section does
not actually represent the text section.
2. Rename the second ROM section, which was previously named
`_TEXT_SECTION_NAME_2` which supposedly refers to the definition of
the same name that does not exist, to `_TEXT_SECTION_NAME`. Note that
this is indeed the section that contains the text section from the
source image.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>