Make the GNSS emul driver less capable w.r.t fix interval so that we can
exercise more error paths if e.g. user requests a fix interval of 100ms.
Signed-off-by: Benjamin Cabé <benjamin@zephyrproject.org>
arch_elf_relocate_global() and arch_elf_relocate_local() aren't LLEXT
API functions, they're internal implementation functions, move them
to llext_internal.h
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
With experience it becomes clear, that failing to resolve symbols
during the linking process is likely fatal for the module loading and
a simple warning isn't enough. Fail loading instead.
Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Functions like `add` and `sub` can only return base 10 integers, which
means they can't really be used to define Kconfig symbols of type `hex`.
For the same reason, there already exist pairs of devicetree functions
named e.g., `dt_node_reg_addr_(int|hex)` after different return types.
Introduce `add_hex`, `sub_hex`, and friends.
To avoid confusion, it should be possible for those new functions to
accept arguments in base 16 as well. It's actually easier to let all
arithmetic functions take their inputs in "any" base, by leveraging
Python's built-in: `int(..., base=0)`.
Signed-off-by: Grzegorz Swiderski <grzegorz.swiderski@nordicsemi.no>
The hci_ipc sample does not support fragmentation of IPC-messages.
Give an ASSERT if only parts of an IPC-message is delivered.
This would happen for instance when the underlying transport mechanism
cannot hold the complete message.
Background:
The open-amp backend uses RPMSG_BUFFER_SIZE=512
Most applications and controllers use HCI-packets which are at most
255 bytes. In some cases a controller may deliver HCI-packets beyond
255 bytes.
This may for instance be the case if the controller delivers large ISO
SDUs as a single HCI-packet.
Signed-off-by: Johan Stridkvist <johan.stridkvist@nordicsemi.no>
The "net conn" command printed AF_PACKET socket as AF_UNK(3).
Fix this by printing "AF_PACKET" in this case.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
Make sure that uninitialized state is 0 so that we can catch
when the context is not initialized properly.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
Make sure we cleanup only those DNS servers that belong to
certain network interface when the interface goes down.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
Bind the network interface to the network interface we
have received the DNS servers from. This is now the default.
The previous behavior can be restored by disabling the
CONFIG_NET_DHCPV6_DNS_SERVER_VIA_INTERFACE option.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
Bind the network interface to the network interface we
have received the DNS servers from. This is now the default.
The previous behavior can be restored by disabling the
CONFIG_NET_DHCPV4_DNS_SERVER_VIA_INTERFACE option.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
Allow user to specify which network interfaces to bind the
server to when reconfiguring the DNS system.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
Add option in enum phy_link_speed to disable auto-negotiation.
This allows PHY drivers to support disabling auto-negotiation.
Signed-off-by: Fin Maaß <f.maass@vogl-electronic.com>
This commit fixes infinite loop reported in
https://github.com/zephyrproject-rtos/zephyr/issues/91242, not deallocating
array with virtqueues and not deactivating the virtqueues in case of
failure
Signed-off-by: Jakub Michalski <jmichalski@antmicro.com>
Add support for setting RGMII RX and TX internal delays via DT properties:
`ti,rx-internal-delay` and `ti,tx-internal-delay`.
Signed-off-by: Venkatesh Odela <venkatesh.odela@amd.com>
is_area_readable is used for SOCS other than LPC55XXX chips, not just
chips other than LPC55S36. Change the condition which elides this code
to avoid a GCC 14.3 warning.
Signed-off-by: Keith Packard <keithp@keithp.com>
In the previous code, strnlen could have returned WIFI_SSID_MAX_LEN, and
the following statement ensuring NUL termination would have written one
past the end of the array.
Replace this with code that ensures a NUL termination within the bounds
of the array and then use strlen to compute the length.
Signed-off-by: Keith Packard <keithp@keithp.com>
We want the stack trace to show the correct nesting of functions,
so we need to prevent the compiler from in-lining functions or
performing tail-call optimizations.
Add __noinline to prevent inlining. Add another printf in each
function to avoid tail calls.
Signed-off-by: Keith Packard <keithp@keithp.com>
GCC 14.3 is not happy when this variable is left uninitialized, although
we don't actually care about the contents, just the address.
Signed-off-by: Keith Packard <keithp@keithp.com>
SUBALIGN forces alignment to the specified value, even if the object
requires stricter alignment. This causes mis-aligned access when accessing
the resulting value.
Signed-off-by: Keith Packard <keithp@keithp.com>
Most uses of this macro should *not* include SUBALIGN as that can reduce
the alignment requirement for data and cause unaligned access faults in
the processor. Leave support for this in place in case there are
still uses for it.
Signed-off-by: Keith Packard <keithp@keithp.com>
SUBALIGN forces alignment of input sections to the specified value, even
if they require stricter alignment.
Signed-off-by: Keith Packard <keithp@keithp.com>
inference_process.hpp depended on the C library implicitly including
stdint.h through some path, which is not defined behavior. Explicitly
include <cstdint> to ensure that uint8_t is defined.
Signed-off-by: Keith Packard <keithp@keithp.com>
