* Added mechanism enabling CPU to
enter WFI mode when the OS is
in idle, what results in large
power consumption reduction.
* Added mechanism to switch SDRAM to
self-refresh mode before entering
WFI, what resulted in further power
consumption reduction.
If the previous refresh failed the driver tried to add a new frame
in the front of the vector. In some cases, the updateFrames could
be empty, and adding a frame caused a system crash.
* Adding document describing how to use logger
* Adjusting logs to follow a new guide
* Change order in log header:
line number is now before function name
1. Implement partial refresh.
2. Implement refresh canceling mechanism.
3. Refactor some parts of the gui and display code.
ad 1.
- Detect parts of the screen changed since last update and merge them
into bigger regions. These regions defines parts of the context sent
to the display.
- Refresh the region covering all of the parts since this is the most
time consuming part and the size of the refreshed region doesn't
change the time much.
- Refresh the whole screen if deep refresh is requested and previously
fast refresh was used. This is needed to prevent unwanted artifacts
in some cases.
ad 2.
- Separate display update and refresh logic.
- Divide image display message handling into two handlers, one updating
and other one refreshing the screen.
- Add cancel refresh message and use it to cancel refresh during update.
- Store sum of refresh regions gathered during updates to refresh them
all at once at the end.
Detect parts of the screen changed since last update and merge them into
bigger regions. These regions defines parts of the context sent to the
display.
Refresh the region covering all of the parts since this is the most time
consuming part and the size of the refreshed region doesn't change the
time much.
Refresh the whole screen if deep refresh is requested and previously
fast refresh was used. This is needed to prevent unwanted artifacts in
some cases.
Refactor some parts of the gui and display code.
While restoring system data the timer can turn off power of E-ink
to save the power. To avoid this situation we need to power on
the E-ink before wiping out the display. The restore procedure
also performs reversing services list to close. In this case
we do it twice so the order is wrong.
* Dependencies between service-gui and service-eink
have been correctly defined.
* System close routine was replaced with the sequential one
which should solve many bell issues.
* It also fixes BH-1406 and BH-1442.
* Fixed some weird CMake issues in module-audio and
several other modules.
* Fixed service-gui closing routine
* Removed unneeded messages/states from service-gui/eink.
* Enhanced app-mgr closing routine. Fixed closing action handling.
It appears in the past a major edit was made but had used a regex to
change some text which resulted in an errant space between the
enumuneration/struct/class name and the scope resolution operator.
These errant spaces have been removed.
Added some services specific implementations of pre closure procedure.
If one request pre closre, all mesagges (except system ones) recieved
by service are ignored.
Improve structure of the module sys. Problems within this module
prevents from linking others.
Signed-off-by: Marcin Smoczyński <smoczynski.marcin@gmail.com>
Split module-utils into smaller components for better dependency
management. Fix many dependencies throughout the project.
Signed-off-by: Marcin Smoczyński <smoczynski.marcin@gmail.com>
Increase stack sizes of the eink service and the system watchdog. During
stability tests stack-overflows were observed for both threads.
Signed-off-by: Marcin Smoczyński <smoczynski.marcin@gmail.com>
(cherry picked from commit f9a23a7b58)
In order to synchronize the Low Power mode, the services were
immediately informed about the frequency change,
so that they can update their resources (e.g. PWM filling)
and services may request the maximum CPU frequency in order
to perform a task (e.g. screen redraw, telephone conversation)
Do not waste time right before refresh of the eink
display to measure new temperature. Assume that
the measurement from early frame prepare stage is valid.
It has only to be valid for only <time of rendering the frame>,
which is less than a 1000ms.
Due to vfs deprecation there is need to remove all vfs calls from code. This PR covers module services. It also contains minor fixes in module-services.
* EGD-3585 SerwiceGUI and ServiceEink initial cleanup
* moved code to one function per message handling
* removed dead code from comments
* added Service.cpp demangling to debug messages for clear output
* added Response value to return messages to `connect()`
