Putting data to local buffer before transmission was removed
to optimize operation. Local buffering was not needed, as passed
buffer cannot be modified until sending is finished.
Signed-off-by: Kamil Kasperczyk <kamil.kasperczyk@nordicsemi.no>
Currently, functions for poll sock_fds array management are buggy, in
case there is another socket open (for instance the socket for firmware
update download), it could get overwritten, if the LwM2M socket was
closed and re-opened in a meantime (e. g. on registration timeout).
Fix this, by appending new entries to the sock_fds in continuous manner.
In case of removal, the deleted entry is overwritten by the last one,
and the last one is cleared.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
During FW update, the application expects a consecutive data stream.
Therefore retransmitted blocks shall not be forwarded to the
application, but ignored. In case blocks are received out of order,
return an error and do not handle this block.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
When FW update in PUSH mode is used, the firmware is encapsulated in the
TLV as an opaque data, according to the LMWM2M satandard, and then
sliced into blocks and transferred block by block in several
transactions. Therefore, the TLV header is only present in the initial
message.
Current implementation did not handle this case well, reporting errors
on consecutive blocks, therefore making the FW update in PUSH mode
broken.
This commit fixes this issue with following changes:
* The TLV is only assumed to be present in the initial block, while
consecutive blocks will be processed directly into the appropriate
handler,
* 32-bit variables shall be used whenever dealing with the opaque data
length, since the firmware size can easily exceed the 16-bit range,
* Additional information, required for the FW block transfer to work
properly were added to the block context structure,
* The application shall only be notified of the actual data length, and
not the total block size (the total TLV size including header).
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
When the dupplicate address detection mechanism is not used, we should
not wait for IPv6 address configuration. Because of the bug we are
waiting for it, when CONFIG_NET_CONFIG_MY_IVP6_ADDR is not set.
This commit fixes it.
Signed-off-by: Lukasz Maciejonczyk <lukasz.maciejonczyk@nordicsemi.no>
Remove the `sm_bootstrap_reg_done` function, which produced duplicated
logs. The bootstrap registration done event is already logged on state
transition. Additionally, in case bootstrap procedure took longer time,
the duplicated log message was printed on each `lwm2m_rd_client_service`
call (500 miliseconds by default).
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
The security object instance used should be cleared on engine reset,
otherwise we might end up using invalid object instance for the
registration (i. e. if the engine was restarted during the bootstrap
procedure due to socket errors, the registration attempt will use
bootstrap security object instead of finding a proper one).
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
During the bootstrap procedure, when Boostrap Finish was received, the
response message was not initialized properly, resulting in a socket
error (NULL pointer porovided) and the response not being sent.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Fix compilation error when the DIAG module is enabled.
All core functionalities are provided.
Signed-off-by: Piotr Szkotak <piotr.szkotak@nordicsemi.no>
`addrlen` parameter is updated on each `recvfrom` call, indicating the
actual address length returned. In case both, IPv4 and IPv6 are used on
different sockets (i. e. on regular LWM2M socket and FOTA socket), the
returned address length will differ.
In case `from_addr_len` is not reinitialized on each iteration, the
value stored in the `from_addr_len` variable will eventually indicate
the smaller IPv4 address size, therefore resulting in a failure in a
consecutive call on an IPv6 socket.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Although LwM2M engine uses cooperative threads, the internal `send()`
implementation might trigger context switch when it calls a kernel
function, therefore resulting in `send()` call being entered from both
the LwM2M thread and the retransmit work.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Unit tests were failing to build because random header was included by
kernel_includes.h. The problem is that rand32.h includes a generated
file that is either not generated or not included when building unit
tests. Also, it is better to limit the scope of this file to where it is
used.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
The context should only clear messages it owns, not all of them. Since
both context (LwM2M and FOTA) share common message pool, they might
interrupt their operation otherwise (i. e. cancel retransmissions).
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
The FOTA socket was not closed when download finished or an error
occured.
Additionally, fix the socket fd verification (it was assumed 0 is not a
valid fd which is not correct).
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
In case unsopported TLV type or malformed packet is received, the
`do_write_op_tlv` function will end up in an infinite loop. Prevent that
by returning an error code in case it does not recognize TLV type.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Allow to configure, how long before registration timeout should the
Registration Update be sent. The fixed 6 seconds used so far, might
not be enough in slower networks (like NB-IoT), resulting in frequent
re-registrations at LWM2M level.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
OpenThread API is not thread-safe.
Moved creation of otMessage to the Thread task and created api
for passing it properly.
This way it should be less possible for an issue to occure eg.
during message buffer allocation.
Signed-off-by: Marek Porwisz <marek.porwisz@nordicsemi.no>
Added automatic strdup for RAM %s parameters. Postponed format
processing to idle time. Automatic parameters counting during
compilation. Very fast execution for up to three parameters.
Signed-off-by: Tomasz Konieczny <tomasz.konieczny@nordicsemi.no>
This is no longer required since drivers implementing the sockets
offload interface were migrated to use pure zsock_ instead of
raw POSIX types and functions.
Signed-off-by: Adam Porter <porter.adam@gmail.com>
With timeout==0 count is set to 0 as well and then it is decremented to
-1. It is later checked == 0 and continues to decrement and loop.
Signed-off-by: Markus Becker <markus.becker@tridonic.com>
Under Linux when you shutdown a socket which is blocked on
an accept call the error code returned by accept is EINVAL.
Modify the socket code to be inline with this behaviour.
Signed-off-by: Léonard Bise <leonard.bise@gmail.com>
When closing a listening socket the functions waiting on the
FIFO will be unblocked this will result in receiving a NULL child
context. If that is the case return an error instead of carrying on.
Return the same error code (EINVAL) that is returned on Linux when
calling shutdown on a blocked accept call.
Signed-off-by: Léonard Bise <leonard.bise@gmail.com>
In case CoAP packet does not carry payload, and the last option is
zero-length, the option structure was not filled with data (opt number
and length). Fix the length check to prevent this from happening.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
This should be boolean according to specification. It makes difference
when JSON encoding is used.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
So far net config initialization code relied on number of services
needed by application. This was fine as long as all enabled
services (e.g. CONFIG_NET_IPV4=y, CONFIG_NET_IPV6=y) were selected by
project configuration as "needed" (e.g. CONFIG_NET_CONFIG_NEED_IPV4=y,
CONFIG_NET_CONFIG_NEED_IPV6=y). Problem appeared for example when both
IPv4 and IPv6 were enabled (CONFIG_NET_IPV4=y, CONFIG_NET_IPV6=y), but
only IPv6 was marked as "needed" (CONFIG_NET_CONFIG_NEED_IPV6=y). In
such situation number of required services was equal to 1. When IPv4
setup was completed, this resulted in returning from net_config_init()
function. Application code failed, because IPv6 was still not
functional.
Do not rely on number of services anymore, as it is error prone. Use
flags instead to mark which services are ready. Compare those flags with
the flags passed to net_config_init() (selected in project configuration
in most cases) to decide whether network configuration has completed
already or not.
Also introduce services_notify_ready() and services_are_ready() helper
functions to isolate implementation details from the logic.
Signed-off-by: Marcin Niestroj <m.niestroj@grinn-global.com>
With current design there is single semaphore (called 'waiter') for
wakeing up initialization thread. This thread should then check for
number of protocols that still need to be initialized. This happens now
only when waiting on 'waiter' semaphore times out.
Do not check for k_sem_take(&waiter) return value, as all needed
information about protocols being initialized already is in 'counter'
semaphore.
Fixes: #25358
Signed-off-by: Marcin Niestroj <m.niestroj@grinn-global.com>
It might happen that while we are waiting for the response
to one of our query, we receive a query to resolve another name.
Previously this would make the current name resolution to fail because
only responses were expected to be received.
Signed-off-by: Léonard Bise <leonard.bise@gmail.com>
Any CoAP implementation when use at least block transfer or is a server
side need access some CoAP options as integer values. This add a method
at public interface and defines for block wise operations to avoid code
useless code duplication.
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
The SET_ERRNO() macro does nothing if a positive value is provided
to it, and the functions were not returning -1 or setting errno
as expected.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
APIs were returning -1 without setting errno if the file
descriptor looked up a null object or there was no function
installed in the vtable. Set to EBADF for this case.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Cleansession is hardcoded to 1 but some use case might require 0
to make the sessions persistent (ie get messages sent while the
client was offline)
Signed-off-by: Xavier Naveira <xnaveira@gmail.com>
This PR enables the user to configure the maximum command length for
OpenThread's CLI using Zephyr's shell, which is currently fixed to
256 characteres. This limitation was not enough for certain allowed
OpenThread commands, like some variants of 'ot dataset mgmtsetcommand'.
Signed-off-by: Eduardo Montoya <eduardo.montoya@nordicsemi.no>
When `z_get_fd_obj_and_vtable()` function returns NULL (no valid entry
in the FD table for the socket), there is no need for further usermode
checks on the `ctx` pointer, as there is nothing to invalidate in that
case.
Fixes#25990Fixes#25991
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
The original sockets system calls used file descriptors which
were actually net_context pointers. For all socket system calls,
any calls from user mode would check if the caller had permission
to use the net context.
This was later changed to not stuff net_context pointers into file
descriptors, but all the permission checking was unintentionally
lost, allowing all threads on the system to
read/write all socket file descriptors in the system at will, with
no way to isolate applications running on the same microcontroller
from each other's network activity.
This patch restores the permission checks on network context objects
for socket system calls that originated from user mode.
The call to z_object_recycle() was never removed from
zsock_socket_internal(); this is again leveraged to grant the
caller who opened the socket permission on the net_context
associated with the returned file descriptor.
To ensure that all socket calls do this checking, all uses of
z_get_fd_obj_and_vtable() have been routed through get_sock_vtable().
Objects have initialization state set and thread permissions
reset to just the caller in common zsock_socket() code.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The socketpair file descriptor context objects are heap allocated
and not drawn from a static pool. Register these as kernel objects
when we create them if user mode is enabled.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Zephyr running on MPU devices have a different memory model than
process-oriented OSes like Linux and require a method to set
kernel object permissions on a file descriptor's underlying
context object. Add this, and a test to show that it is working.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
