455507 - what to do about browserd

Status: VERIFIED INVALID

(Firefox for Android Graveyard :: General, defect)

Description

[Doug Turner (:dougt)]

browserd eat a ton of memory that fennec would be very happy have.  Maybe we should stop the browserd at install time or something?

Comment 1

[Oleg Romashin (:romaxa)]

if [ "" == "$(ps | grep '/usr/bin/browser' | sed '/grep/d')" ];then
kill `ps | grep "browserd -s" | sed '/grep/d' | awk '{print $1}'`;
fi

Comment 2

[Oleg Romashin (:romaxa)]

#1 command will kill browserd process if browser-ui is not running...

We can do it also every time before fennec running

Comment 3

[Doug Turner (:dougt)]

there is also the service we can kill too, right?  

/etc/init.d/tablet-browser-daemon stop

Comment 4

[Oleg Romashin (:romaxa)]

(In reply to comment #3)
> there is also the service we can kill too, right?  
> 
> /etc/init.d/tablet-browser-daemon stop
Right, but after that user not able to do /etc/init.d/tablet-browser-daemon start without terminal cmd... and not able to use default browser (until reboot).

Comment 5

[timeless]

absolutely not.

if you replace the bookmarks manager item in the task navigator, then maybe. but unless install time also removes the related browser things.

Comment 6

[Doug Turner (:dougt)]

timeless is probably right.  this is a "bug" in Maemo.  we should take it up with them.  

Someone will probably just provide a fennec extension does the dirty workaround.

Comment 7

[Antonio Gomes (tonikitoo)]

people tried to elaborate on that w/out success

https://bugs.maemo.org/show_bug.cgi?id=3705

Comment 8

[timeless]

that wasn't a good bug report.

this is different. it is theoretically possible to replace the browser.

you'd want to provide a component which provides certain things and conflicts with certain things. that'd cause the browser to be uninstalled.

note that the browser is a service, it's used by other clients.

this is similar to how MSHTML works on windows.

If you actually cause the browserd itself to be uninstalled, then you *must* provide a replacement (which would be a dbus compatible daemon).

various apps on our *platform* use the browser daemon including the tablet tutorial.

fwiw, i think we were supposed to have worked on the browser-ui such that it could start the daemon if it wasn't already running, so it may be possible to arrange to stop the browser when fennec starts if there's no consumer.

Comment 9

[Doug Turner (:dougt)]

food for thought -- a service/singleton could be started/created when it is first needed.  This would allow more memory for all existing application that do not require such a service.  You could even be smart with the service and shut it down after all users go away.

Comment 10

[Doug Turner (:dougt)]

food for thought -- a service/singleton could be started/created when it is
first needed.  This would allow more memory for all existing application that
do not require such a service.  You could even be smart with the service and
shut it down after all users go away.

Comment 11

[timeless]

yeah, we know. this was a design choice. it sped up the average case.

i think what we [maemo software/browser team] probably should do is kill the browser instance if there's no ui and we get a low memory signal. that should solve the problem for the only case that matters.

Comment 12

[Doug Turner (:dougt)]

sounds reasonable.

could we make sure that if there is no ui, that browserd gets the signal before other applications?  not sure if this is possible or not, but it would allow other applications not to panic during this OOM situation (browserd would exit first freeing up some memory possibility saving other applications from getting the OOM signal)

Comment 13

[timeless]

sadly, i don't think it's possible like that.

otoh,... there are two watermarks, and i don't know which watermark is responsible for killing. the daemon could probably listen for the lower watermark and quit then....

Comment 14

[Mark Finkle (:mfinkle) (use needinfo?)]

I don't think we are going to do anything at this time.

Comment 15

[Biju]

sorry for bug spam.
Many of the bugs which are marked invalid, I see comments telling it occurred in one version or other. But later it was fixed due to 1) by backingout the patch which made regression or 2) by fixing some other bug.
So if we can identify the bug/patch/reason then we should state that and mark those as FIXED. If not mark as WORKSFORME in that case.