Persistence Client Library Help
justin at ics.com
Mon Apr 3 11:58:50 EDT 2017
There is an old Ubuntu bug about something similar where systemd was
not sending a polite TERM before KILL. There is a minimal test program
attached as a comment on the bug report which could be useful to test
the systemd behavior on GDP. See if that program successfully writes a
file with a time stamp in it.
On Mon, Apr 3, 2017 at 11:49 AM, Vy Duong <vduong at ics.com> wrote:
> It does get saved if I stop the application from Qt Creator.
> On Mon, Apr 3, 2017 at 11:47 AM, Justin Noel <justin at ics.com> wrote:
>> I think I can shed some light on the above.
>> 1) The HMI (launcher) exiting will not close any applications. It will
>> happily restart and reclaim the surfaces of the already running
>> applications and continue normally.
>> 2) There UnixSignalHandler class from the HMI is a way to gracefully
>> shut down on SIGTERM and SIGINT signals sent to the process. For
>> example CTRL-C on a termanial would produce SIGINT. Qt Creator sends
>> SIGTERM followed by SIGKILL. Systemd should also send TERM followed by
>> KILL if the process refused to politely shutdown. KILL is not
>> 3) From the description above it sounds like you can save persisted
>> data to disk when your application "cleanly shuts down", but not if
>> it's forced to shut down. Does the persisted data get saved when
>> stopping the application from Qt Creator? That should trigger a
>> SIGTERM via the UnixSignalHandler.
>> On Mon, Apr 3, 2017 at 10:16 AM, Vy Duong <vduong at ics.com> wrote:
>> > Hi,
>> > I am writing an application that makes use of the PCL. In order for
>> > information to be retained across boots, my application needs to do some
>> > "cleaning up" in terms of de-initializing the library and having NSM do
>> > the
>> > actual writing of the data to the device or set the life cycle.
>> > This email thread has the summary of how the library works:
>> > https://lists.linuxfoundation.org/pipermail/genivi-persistence/2016-April/000256.html
>> > Initially, I had relied on the NSM to do the final writing, but this
>> > didn't
>> > work since the NSM doesn't get kill before the application does and
>> > there is
>> > no real "quitting" of either applications. So, I had gone to the latter
>> > option, write back controlled by application.
>> > I had noticed that the HMI code has a "UnixSignalHandler" class that
>> > handles
>> > when the application closes based on the interrupt or termination of the
>> > application. I copied those files into my application folder and was
>> > able to
>> > do connect the quit of the application to the clean up code.
>> > A follow up problem I am experiencing right now is that the HMI may quit
>> > or
>> > terminate without notifying the underlying applications that it has
>> > exited.
>> > Can you please provide some advice on whether including the
>> > UnixSignalHandler was the way to go or if there is an alternative
>> > option?
>> > And thoughts on how to handle when the HMI quits, but the application
>> > does
>> > not recognize that it has exited?
>> > Thank you very much in advance!
>> > Best,
>> > Vy
>> > --
>> > Vy Duong, Software Engineer, Integrated Computer Solutions.
>> > ICS - Delivering World-Class Applications for Embedded & Mobile Devices
>> > http://ics.com/services/
>> > _______________________________________________
>> > genivi-projects mailing list
>> > genivi-projects at lists.genivi.org
>> > http://lists.genivi.org/cgi-bin/mailman/listinfo/genivi-projects
> Vy Duong, Software Engineer, Integrated Computer Solutions.
> ICS - Delivering World-Class Applications for Embedded & Mobile Devices
More information about the genivi-projects