When LOG_MODE_IMMEDIATE is used, logs processed by the shell
log backend may be intertwined with messages printed by
shell commands running on the shell thread.
It is because the shell uses a mutex while the shell log
backend uses the IRQ lock for synchronization. Switch the
latter to use the mutex as well whenever it's possible.
Signed-off-by: Damian Krolik <damian.krolik@nordicsemi.no>
Renaming objects which had 2 in the name to indicate that
it is v2 specific. Once logging v1 has been removed such
suffixes are redundant.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
If CONFIG_SHELL_STATS is disabled, shell->stats is NULL and
must not be dereferenced. Guard against it.
Fixes#44089
Signed-off-by: Alexej Rempel <Alexej.Rempel@de.eckerle-gruppe.com>
In order to bring consistency in-tree, migrate all subsystems code to
the new prefix <zephyr/...>. Note that the conversion has been scripted,
refer to zephyrproject-rtos#45388 for more details.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Reduced logging mode selection to deferred, immediate, minimal and
frontend. Decoupled logging version from mode and created CONFIG_LOG1
which can be used to explicitly select deprecated version.
From now on, chosing CONFIG_LOG_MODE_{IMMEDIATE,DEFERRED} will result
in version2.
Deprecated CONFIG_LOG2_MODE_{IMMEDIATE,DEFERRED} with cmake warning.
Codebase adapted to those changes.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
When logging is using immediate mode then logging messages
can be processed from any context, including interrupt context.
z_shell_fprintf was asserting in that case since it allowed to
be called from interrupt context only when logging was in panic
mode. However, shell works in the same way when logging is in
immediate mode as in panic mode.
Renamed internal shell flag from panic_mode to sync_mode. Flag
is also set when shell log backend is started in synchronous
mode (immediate logging) which prevents assertion.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
In panic mode, the function: z_shell_fprintf is expected to be
called from an interrupt context. Therefore, the dedicated assert
cannot be checked in this case.
Fixes#38612
Signed-off-by: Jakub Rzeszutko <jakub.rzeszutko@nordicsemi.no>
When shell log backend was disabled fifo used only in
deferred mode was unconditionally flushed which lead to
errors in immediate mode where fifo was not present. On
the other hand, in case of LOG2_MODE_DEFERRED fifo (mpsc_pbuf)
was initialized in enable stage.
In order to clean things up, deferred v1 fifo flushing and
deferred v2 mpsc pbuf initialization were moved to fifo_reset
function which is called when shell log backend is enabled.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
When shell had too many pending log messages it was attempting to
drop expired messages and retrying to put the new message. There
was an assumption that enough messages are dropped and new message
can be put. It may not be the case if no message expired during
given time. Wrapped the operation in loop to continue until expired
message is freed and new message is enqueued.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
-Wimplicit-fallthrough=2 requires a fallthrough comment or a compiler
to tells gcc that this happens intentionally.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Kernel timeouts have always been a 32 bit integer despite the
existence of generation macros, and existing code has been
inconsistent about using them. Upcoming commits are going to make the
timeout arguments opaque, so fix things up to be rigorously correct.
Changes include:
+ Adding a K_TIMEOUT_EQ() macro for code that needs to compare timeout
values for equality (e.g. with K_FOREVER or K_NO_WAIT).
+ Adding a k_msleep() synonym for k_sleep() which can continue to take
integral arguments as k_sleep() moves away to timeout arguments.
+ Pervasively using the K_MSEC(), K_SECONDS(), et. al. macros to
generate timeout arguments.
+ Removing the usage of K_NO_WAIT as the final argument to
K_THREAD_DEFINE(). This is just a count of milliseconds and we need
to use a zero.
This patch include no logic changes and should not affect generated
code at all.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The variable needs to be an atomic_t, and in one case it was
being incremented outside of an atomic_inc/atomic_add.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
With CONFIG_LOG_IMMEDIATE, log output is printed immediately. If a
shell command is in progress, there's no prompt to erase, nor should
we print a new prompt after the log message is output.
Before this patch, a simple shell command like:
int cmd_log_erase(const struct shell *shell, size_t argc, char **argv)
{
LOG_INF("hello world");
return 0;
}
would output something like:
uart:~$ log erase
[00:00:02.623,718] <inf> cmd_log: hello world
uart:~$ loguart:~$
This patch fixes prompt handling while a command is active, and fixes
put_sync_hexdump to behave like put_sync_string.
Signed-off-by: Jim Paris <jim@jtan.com>
Extended shell to be able to process logs in place
(in the context of a log call). In order to achieve that,
shell was extended to support for TX blocking operations. If
CONFIG_LOG_INPLACE_PROCESS is enabled then shell instance
attempts to be initialized in blocking TX mode. If fails to
do so, shell log backend is disabled. If successfully enabled
logs are processed and printed in the context of the call.
Due to that change, user may expirience interleaved output as
shell has no means to multiplex shell output with logger output.
In extreme, huge amount of log messages may prevent shell thread
execution and shell may become unresponsive.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Log message get timestamp when being added to shell log message queue.
When adding to log message queue timeouts then all messages added
before timeout are dropped.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
If burst of log messages was passed to the shell log
backend, it was likely that messages were lost because
shell had no means to control arrivals of log messages.
Added log message enqueueing timeout to the shell instance
to allow blocking logger thread if short-term arrival rate
exceeded shell capabilities.
Added kconfig option for setting log message queue size
and timeout in RTT and UART instances. Added section in
shell documentation which explains interaction between
the logger and shell instance acting as a logger backend.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Shell log backend was using k_fifo to enqueue log messages.
It was using field in log message that was used for same
purpose in log_core before passing message to backends.
However, this method supported only single shell as
other shell was corruption the fifo because field was
reused.
Modified shell log backend to use k_msgq for pending
messages.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
k_poll_signal was being used by both, struct and function. Besides
this being extremely error prone it is also a MISRA-C violation.
Changing the function to contain a verb, since it performs an action
and the struct will be a noun. This pattern must be formalized and
followed and across the project.
MISRA-C rules 5.7 and 5.9
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Added flag in log_output module to add timestamp when message is
formatted to a string. Updated existing backends.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Added flag in log_output module to add severity level when message is
formatted to a string. Updated existing backends.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Damian Krolik