MISRA C:2012 Rule 14.4 (The controlling expression of an if statement
and the controlling expression of an iteration-statement shall have
essentially Boolean type.)
Use `do { ... } while (false)' instead of `do { ... } while (0)'.
Use comparisons with zero instead of implicitly testing integers.
The commit is a subset of the original auditable-branch commit:
5d02614e34a86b549c7707d3d9f0984bc3a5f22a
Signed-off-by: Simon Hein <SHein@baumer.com>
If TCP server closed notification is received it
could interrupt the TCP socket RX or TX data sequence.
Remove the SOCK_SERVER_CLOSED state and instead
use a unique error code to know if the socket
has been closed by the server.
Signed-off-by: Ryan Erickson <ryan.erickson@lairdconnect.com>
There is is an error in the way that the sock id is
determined. A unique fd is reserved and assigned appropriatly, but
the id, which should correspond to a socket number within the modem,
is set to an invalid, and identical for all sockets, value, and then
not used appropriatly in the drivers, instead, the sock_fd is used,
which can be any value really, it is not related to the socket number
in any way.
This results in the drivers only working if the reserved fd happens
to be between base_socket_num and (socket_len - 1)
This patch assignes the id to the index of the socket + the
base_socket_num, the socket at index 0 will get the id 1 if the
base_socket_num is 1, and the modem_socket_from_id should then
be used to get a pointer to the socket, since the id is not
neccesarily equal to the index.
The FIXME has been solved by adding a note both at the start
of the modem_socket_get function and inside the Kconfig file
for the MODEM_SOCKET option description. It is not an error,
but the user must be aware that it uses the POSIX file
descriptors, for which only 4 are allocated by default.
This patch fixes the bug, without breaking the brittle modem
drivers which currently are built around this bug.
The modem drivers should be updated to use the id as the
socket num instead of the sock_fd after this fix has been merged.
The "socket # needs assigning" has been removed, as that is what
this patch is doing
I also added comments to the id and sock_fd in the modem_socket
structure to help developers use the id and fd appropriately.
Signed-off-by: Bjarki Arge Andreasen <bjarkix123@gmail.com>
Simplify the implementation by using gpio_dt_spec.
As part of this use logical values of 0/1 for gpio_pin_set_dt
and handle any inversions as GPIO flags that might be required
at a board level (in dts config).
Signed-off-by: Kumar Gala <galak@kernel.org>
Use logical values of 0/1 for gpio_pin_set_dt and handle any
inversions as GPIO flags that might be required at a board level.
Signed-off-by: Kumar Gala <galak@kernel.org>
Following zephyr's style guideline, all if statements, including single
line statements shall have braces.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Change logic in `modem_cmd_handler_setup_cmds` and
`modem_cmd_handler_setup_cmds_nolock` to always sleep
after each sent AT command.
The issue was found using a logic analyzer while debugging 'NO CARRIER'
from a BG95 modem while being in data mode.
The reason 'NO CARRIER' occured was due to crosstalk to DTR which
terminated data mode.
The logic analyzer also revealed that without the delay the next AT
command is sent before the whole 'OK\r\n' is received. This is not
addressed in this commit.
Fixes https://github.com/zephyrproject-rtos/zephyr/issues/47082
Signed-off-by: Jeppe Odgaard <jeppe.odgaard@prevas.dk>
If hw-flow-control is set in the uart section of the device tree source,
it is read by gsm_ppp.c and AT+IFC=2,2 is sent to the modem.
Fixes https://github.com/zephyrproject-rtos/zephyr/issues/46928.
Signed-off-by: Jeppe Odgaard <jeppe.odgaard@prevas.dk>
Default mode must be debug or it can't ever be enabled.
Allow debug log messages to be printed when using
mdm_hl7800_send_at_cmd API.
Add logging to active bands and network coverage commands.
Signed-off-by: Ryan Erickson <ryan.erickson@lairdconnect.com>
Using a global error flag could lead to an error on one
socket appearing as an error with a separate socket instance.
Socket errors moved into the socket context.
Continue to use global error flag for commands not
related to sockets.
For TCP sockets closed by the server, don't try and close
the socket again when offload_put is called.
Signed-off-by: Ryan Erickson <ryan.erickson@lairdconnect.com>
Logging v1 has been removed and log_strdup wrapper function is no
longer needed. Removing the function and its use in the tree.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Files including <zephyr/kernel.h> do not have to include
<zephyr/zephyr.h>, a shim to <zephyr/kernel.h>.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
These drivers use stack buffers to hold AT command strings which are
generated at runtime using sprintk. The buffers are only sized for the
expected range of values, not the full possible range given the datatypes
involved. Values outside this expected range could cause a buffer overflow.
To mitigate this, increase the size of each buffef to hold the full range
of each parameter type.
Signed-off-by: Keith Packard <keithp@keithp.com>
Log message references gsm->context.data_rssi which is NULL
when CONFIG_MODEM_SHELL=n. Use gsm->minfo.mdm_rssi which is
internally used when accessing RSSI value.
Signed-off-by: Marek Metelski <marek@metelski.dev>
Check if ppp iface was actually up before taking it down and waiting
for NET_EVENT_IF_DOWN. This allows to call gsm_ppp_stop() even if
the PPP was never started, which previously caused the call to lock
forever. This case is useful, for example, if the modem didn't attach
to network in desirable time and user wants to abort PPP setup.
Signed-off-by: Marek Metelski <marek@metelski.dev>
Do not truncate outgoing datagrams if they do not fit in maximum handled
packet size. Set EMSGSIZE errno and return -1 instead.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Coverity reported unused value for a variable "ret" in the function
send_socket_data(). Indeed there were two subsequent assignments
to "ret", meaning that the first one was unused.
The root cause of the problem is not checking the value of "ret"
after correctly invoking a function within send_socket_data(). Fix
this by adding appropriate checks.
Coverity-CID: 215249
Signed-off-by: Aleksandar Markovic <aleksandar.markovic.sa@gmail.com>
Check if there are multiple non-empty data fragments passed to sendmsg()
function. If positive, then set EMSGSIZE errno and return -1, as that case
is not handled properly with current implementation.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
If one data fragment was not sent successfully in sendmsg(), then
attempting to send the rest does not make any sense, as it would introduce
hole in data stream.
Currently if one fragment was not sent successfully, then there is a
'break' in inner loop, but that moves the execution to sending the
following data fragments.
Return early in case of fragment sending failure, so that there are no more
send attempts.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
added user_data field to modem_cmd_handler_data
this allows for multiple instances of modem drivers which use
the modem cmd handler. currently, the only identifiable
parameter passed to the modem command handlers is the
modem_cmd_handler_data struct.
The added user_data variable allows for the a modem driver to
pass its dev or data pointer to the modem_cmd_handler_data
struct to be retrieved from within the modem command callbacks.
Signed-off-by: Bjarki AA <baa@trackunit.com>
All drivers using 'modem_pin' abstraction were converted already, so remove
its implementation now.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Move away from 'modem_pin' abstraction as it has not obvious value compared
to generic 'gpio_dt_spec'.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Move away from 'modem_pin' abstraction as it has not obvious value compared
to generic 'gpio_dt_spec'.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Move away from 'modem_pin' abstraction as it has not obvious value compared
to generic 'gpio_dt_spec'.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
mux_enabled check in the gsm_configure is unnecessary since it
is init and scheduled by gsm_ppp_start which means that the
mux must be disabled. The IS_ENABLED(CONFIG_GSM_MUX) check
should be good enough to determine whether or not the
mux_enable function should be ran.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
After the introduction of start/stop & rssi query work, these
additional functions can be invoked from different threads.
Things can quickly become hard to analyze when an interrupt
fires during one of these functions and upon return the
scheduler schedules another thread and run another function.
This PR introduces a simple mutex lock that guarantees thread
safety. Similar implementation can be found in hl7800 driver,
which has lots of public APIs that can be invoked from different
threads.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
Update gsm_ppp.c
Move the rssi query part out from the rssi_handler work item as
an individual function, so that it can be used directly without
referencing the work.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
Adds a communications backend based on the asynchronous UART API,
instead of the interrupt-driven UART API. The primary advantage of this
backend is an improved robustness to dropping bytes under high interrupt
or critical section loads.
Under all loads system efficiency is improved by:
* Reducing the time spent writing out individual bytes.
* Reducing the number of UART interrupts fired.
* Waking up the RX thread much less often.
When utilising this backend over `nordic,nrf-uarte` on a nRF52840, the
baudrate of an esp-at modem could be pushed to at least 921600 without
dropping bytes, compared to a maximum of 230400 with the interrupt API.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Add a choice symbol that is used to select which UART backend to use.
This allows backends that don't use the interrupt API to be implemented.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Allow CTS line to determine UART shutdown for any sleep mode.
This allows lower average current consumption for LITE
HIBERNATE mode.
Signed-off-by: Ryan Erickson <ryan.erickson@lairdconnect.com>
In order to bring consistency in-tree, migrate all drivers to the new
prefix <zephyr/...>. Note that the conversion has been scripted, refer
to #45388 for more details.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Do not query SIM card parameters if the SIM
card is not present.
This shortens the driver initialization time
significantly if a SIM card is not present.
Signed-off-by: Ryan Erickson <ryan.erickson@lairdconnect.com>
Add a separate macro for registering offloaded sockets implementation,
along with information in the structure whether the implementation is
offloaded or not. This allows to differentiate between native and
offloaded socket implementations, which is critical for binding socket
API with an interface.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Instead of keeping a boolean informing whether a network interface is
offloaded at socket layer or not, keep a pointer to a function which
allows to create an offloaded socket. Native interfaces keep this as
NULL, while for offloaded interfaces it allows to connect an offloaded
socket implementation with an interface.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Check if requested socket family, type and protocol are all supported by
the driver, instead of blindly acknowledging every possible variant.
There is explicit support for TCP on top of IPv4 and IPv6.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Check if requested socket family, type and protocol are all supported by
the driver, instead of blindly acknowledging every possible variant.
There is explicit support for UDP and TCP on top of IPv4 and IPv6.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Check if requested socket family, type and protocol are all supported by
the driver, instead of blindly acknowledging every possible variant.
There is explicit support for UDP, TCP and TLS on top of IPv4 and IPv6.
TLS seems to be supported only in 1.2 version, so allow just that
version.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Use NET_SOCKETS_OFFLOAD_PRIORITY instead of
NET_SOCKETS_OFFLOAD_PRIORITY, so that by default offloaded sockets will
be used instead of native sockets.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Use configurable NET_SOCKETS_OFFLOAD_PRIORITY instead of hardcoded value
in the driver itself. This allows to select relative priority of
offloaded TLS versus native TLS when used together with
NET_SOCKETS_TLS_PRIORITY.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
Use configurable NET_SOCKETS_OFFLOAD_PRIORITY instead of hardcoded value
in the driver itself. This allows to select relative priority of
offloaded TLS versus native TLS when used together with
NET_SOCKETS_TLS_PRIORITY.
Drop the build assert, as always prioritizing offloaded TLS over native
TLS should be application developer choice.
Signed-off-by: Marcin Niestroj <m.niestroj@emb.dev>
In 90c6dc5e7f, a change was introduced to
allow modem commands determine if they have enough data or not.
In a situation where some data is missing, the command should return
-EAGAIN and this should lead to another call of the command with more
data.
In this commit, the argument parser was also allowed to return -EAGAIN
to request more data due to missing arguments.
However, this can't work because in cmd_handler_process_rx_buf() before
calling process_cmd():
- we make sure that a CR/LF has been found.
- we compute match_len which can't be greater than the distance to the
next CR/LF.
Therefore, even if the command argument parser ask for more data by
returning -EAGAIN, next call will have the same value for match_len,
meaning that the parsing of argument will result in the same missing
argument situation.
This leads to an infinite loop of parsing the same data over and over in
an infinite loop.
This commit change this behavior to always drop the data in such a
situation. The command will not be answered and will therefore timeout,
but at least, next commands will correctly parse their returned data.
Signed-off-by: Xavier Chapron <xavier.chapron@stimio.fr>
Use the Kconfig macros of MODEM_GSM_RX_STACK_SIZE and
MODEM_GSM_WORKQ_STACK_SIZE directly without another macros.
They are well within the column limit and this make them
more obvious that they are actually configurable by Kconfig.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
