flashplayer plugin problem

Here’s the situation:

Our application allows user to view webpages in a web client widget. A
screensaver mode is incorporated into the application that’ll be activated
after a period of inactivity. Once the application goes into the screensaver
mode, it’ll display a html page containing flash animations. The html page
will refresh over a specific period of time, playing another flash animation
or mpeg movie. Normally, flashplayer will exit itself when the webclient
refresh to another html page that does not contain any flash animation.

However, there are cases where flashplayer never exits. Using the
information from ‘pidin time’, flashplayer is reported to have taken up a
few minutes or even up to a few hrs of utime. During this time, the
application cannot receive any events, similar to a ‘hang’ state. The kernel
is still running in the meantime (cronjobs are being executed correctly).
Once flashplayer is killed, the application will recover from the ‘hang’
state and all the pending events(events that were put to the event queue
during the ‘hang’ state) will be executed all at one shot.

I noticed that npl* files are being created in /tmp whenever a new flash
clip is played and removed when the flashplayer stops playing the clip. When
flashplayer starts having the above abnormal behaviour, npl* files are not
removed from /tmp though new npl* files are created when the html page
refresh to play a new clip. We tried removing the npl* files from /tmp to
see if flashplayer will recover but it did not worked. We have about 50
machines running this application, up till now, only a few machines have
this problem. All machines are running QNX6.0 PatchB.
Does anyone knows what are the possible reasons that will cause flashplayer
to keep running in the background and not exit? And why is the application
unable to receive events when this happens? (flashplayer is not taking up a
lot of memory from what we see)

Thanks!!

We have seen this sporadically with our HomeFusion appliance product as
well.

We wound up with a seperate process that constantly watched all the other
processes.
It was customized to know about all the little problems of all the different
apps running
in the system. It would notice multiple instances of FlashPlayer and kill
them off when it thought
they were no longer needed. Naturally this program would have to be
re-tweaked for every version
of the OS (6.0 vs 6.10 vs ?) because some things are fixed in each round and
others compromised.

I wasn’t really comfortable with this “solution” from a software point of
view, but solutions like this are
always required to meet a productoin deadline and to provide a box to a
consumer that feels reliable and robust.
They never know what goes on inside anyway, nor should they.

\

Jerry Kirk
Nexware Corp. (http://www.nexwarecorp.com)
Software Development for the Embedded World
865.546.9998 x203


“Wen” <xiaowen@axs.com.sg> wrote in message news:9q35um$dds$1@inn.qnx.com

Here’s the situation:

Our application allows user to view webpages in a web client widget. A
screensaver mode is incorporated into the application that’ll be activated
after a period of inactivity. Once the application goes into the
screensaver
mode, it’ll display a html page containing flash animations. The html page
will refresh over a specific period of time, playing another flash
animation
or mpeg movie. Normally, flashplayer will exit itself when the webclient
refresh to another html page that does not contain any flash animation.

However, there are cases where flashplayer never exits. Using the
information from ‘pidin time’, flashplayer is reported to have taken up a
few minutes or even up to a few hrs of utime. During this time, the
application cannot receive any events, similar to a ‘hang’ state. The
kernel
is still running in the meantime (cronjobs are being executed correctly).
Once flashplayer is killed, the application will recover from the ‘hang’
state and all the pending events(events that were put to the event queue
during the ‘hang’ state) will be executed all at one shot.

I noticed that npl* files are being created in /tmp whenever a new flash
clip is played and removed when the flashplayer stops playing the clip.
When
flashplayer starts having the above abnormal behaviour, npl* files are not
removed from /tmp though new npl* files are created when the html page
refresh to play a new clip. We tried removing the npl* files from /tmp to
see if flashplayer will recover but it did not worked. We have about 50
machines running this application, up till now, only a few machines have
this problem. All machines are running QNX6.0 PatchB.
Does anyone knows what are the possible reasons that will cause
flashplayer
to keep running in the background and not exit? And why is the application
unable to receive events when this happens? (flashplayer is not taking up
a
lot of memory from what we see)

Thanks!!

So is this a bug of flashplayer or what?
At least now I know we’re not the only one experiencing this problem. :slight_smile:


“Jerry Kirk” <Jerry.Kirk@Nexwarecorp.com> wrote in message
news:9qendr$fr2$1@inn.qnx.com

We have seen this sporadically with our HomeFusion appliance product as
well.

We wound up with a seperate process that constantly watched all the other
processes.
It was customized to know about all the little problems of all the
different
apps running
in the system. It would notice multiple instances of FlashPlayer and kill
them off when it thought
they were no longer needed. Naturally this program would have to be
re-tweaked for every version
of the OS (6.0 vs 6.10 vs ?) because some things are fixed in each round
and
others compromised.

I wasn’t really comfortable with this “solution” from a software point of
view, but solutions like this are
always required to meet a productoin deadline and to provide a box to a
consumer that feels reliable and robust.
They never know what goes on inside anyway, nor should they.

\

Jerry Kirk
Nexware Corp. (> http://www.nexwarecorp.com> )
Software Development for the Embedded World
865.546.9998 x203


“Wen” <> xiaowen@axs.com.sg> > wrote in message
news:9q35um$dds$> 1@inn.qnx.com> …
Here’s the situation:

Our application allows user to view webpages in a web client widget.
A
screensaver mode is incorporated into the application that’ll be
activated
after a period of inactivity. Once the application goes into the
screensaver
mode, it’ll display a html page containing flash animations. The html
page
will refresh over a specific period of time, playing another flash
animation
or mpeg movie. Normally, flashplayer will exit itself when the webclient
refresh to another html page that does not contain any flash animation.

However, there are cases where flashplayer never exits. Using the
information from ‘pidin time’, flashplayer is reported to have taken up
a
few minutes or even up to a few hrs of utime. During this time, the
application cannot receive any events, similar to a ‘hang’ state. The
kernel
is still running in the meantime (cronjobs are being executed
correctly).
Once flashplayer is killed, the application will recover from the ‘hang’
state and all the pending events(events that were put to the event queue
during the ‘hang’ state) will be executed all at one shot.

I noticed that npl* files are being created in /tmp whenever a new flash
clip is played and removed when the flashplayer stops playing the clip.
When
flashplayer starts having the above abnormal behaviour, npl* files are
not
removed from /tmp though new npl* files are created when the html page
refresh to play a new clip. We tried removing the npl* files from /tmp
to
see if flashplayer will recover but it did not worked. We have about 50
machines running this application, up till now, only a few machines have
this problem. All machines are running QNX6.0 PatchB.
Does anyone knows what are the possible reasons that will cause
flashplayer
to keep running in the background and not exit? And why is the
application
unable to receive events when this happens? (flashplayer is not taking
up
a
lot of memory from what we see)

Thanks!!

\