Bug 142147 - A plug-in sometimes leaves MBP discrete GPU processor enabled after closing a window
Summary: A plug-in sometimes leaves MBP discrete GPU processor enabled after closing a...
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: 528+ (Nightly build)
Hardware: Macintosh Intel OS X 10.10
: P2 Minor
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2015-03-01 09:26 PST by Jim Oase
Modified: 2015-03-09 13:37 PDT (History)
5 users (show)

See Also:


Attachments
Reason the discrete GPU can not be turned off after leaving site that used it. (19.37 KB, image/jpeg)
2015-03-01 09:30 PST, Jim Oase
no flags Details
Indicates the plugin being left active after the site using it is inactive. (18.11 KB, image/jpeg)
2015-03-05 08:58 PST, Jim Oase
no flags Details
Screen shot showing Flash plug-in being activated after waking from sleep. (84.99 KB, image/jpeg)
2015-03-08 13:15 PDT, Jim Oase
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Jim Oase 2015-03-01 09:26:38 PST
When a site is accessed that uses the MacBook Pro discrete GPU exiting that site does not release the discrete GPU.  Leaving the GPU causes excessive battery drain and heat build up.

This site requires a password but is an example... http://www.video.theblaze.com/video/v37708683/0225-cain-conversation
Comment 1 Jim Oase 2015-03-01 09:30:12 PST
Created attachment 247632 [details]
Reason the discrete GPU can not be turned off after leaving site that used it.
Comment 2 Alexey Proskuryakov 2015-03-01 16:42:02 PST
This dialog claims that it's due to a plugin.
Comment 3 Tim Horton 2015-03-01 17:15:57 PST
Don't we keep the plugin process up for a while after last using the plugin? Does it resolve itself after a minute or something?
Comment 4 Jim Oase 2015-03-04 15:48:41 PST
There have been occasions when the discrete GPU is deactivated after closing a site that activates the discrete graphic processor.  

So far the process to accomplish deactivation of the GPU is not repeatable.  Time is a function but the period is variable.  Sometimes over night does not affect the deactivation other times 10 minutes is sufficient.

Causing discrete GPU activation and deactivation by launching and closing iPhoto is consistently immediate.
Comment 5 Alexey Proskuryakov 2015-03-04 20:38:24 PST
If the same plug-in type (e.g. Flash) is in use in any other browser window, that would keep the process alive. 

I think that this should be primarily a bug report against the plugin vendor.
Comment 6 Jim Oase 2015-03-04 21:45:24 PST
(In reply to comment #5)
> If the same plug-in type (e.g. Flash) is in use in any other browser window,
> that would keep the process alive. 
> 
> I think that this should be primarily a bug report against the plugin vendor.

Thanks I will check that out... my only plug in is Flash.
Comment 7 Jim Oase 2015-03-04 21:46:38 PST
(In reply to comment #6)
> (In reply to comment #5)
> > If the same plug-in type (e.g. Flash) is in use in any other browser window,
> > that would keep the process alive. 
> > 
> > I think that this should be primarily a bug report against the plugin vendor.
> 
> Thanks I will check that out... my only plug in is Flash.

Just occurred to me... doesn't have that problem with Safari which also uses Flash.  What is different?
Comment 8 Jim Oase 2015-03-05 08:58:41 PST
Created attachment 247955 [details]
Indicates the plugin being left active after the site using it is inactive.
Comment 9 Jim Oase 2015-03-05 08:59:27 PST
(In reply to comment #5)
> If the same plug-in type (e.g. Flash) is in use in any other browser window,
> that would keep the process alive. 
> 
> I think that this should be primarily a bug report against the plugin vendor.



There is a plug in  com.apple.WebKit.Plugin.64.Deveolopment   being left on after the site using the discrete processor is closed.


Safari doesn't use the plugin and doesn't have the problem.
Comment 10 Jim Oase 2015-03-08 13:14:20 PDT
(In reply to comment #6)
> (In reply to comment #5)
> > If the same plug-in type (e.g. Flash) is in use in any other browser window,
> > that would keep the process alive. 
> > 
> > I think that this should be primarily a bug report against the plugin vendor.
> 
> Thanks I will check that out... my only plug in is Flash.


When returning from sleep the discrete processor is being turned on.  This is after operating with the same windows for a half hour.  None of those windows caused the discrete processor to be activated.

After a minute or two the discrete processor is turned off again.

I will attach a screen shot from the Activity Monitor showing that the Flash Player (SafariForWebKitDevolopment Internet Plug-in) in is being activated and then inactivated.
Comment 11 Jim Oase 2015-03-08 13:15:53 PDT
Created attachment 248198 [details]
Screen shot showing Flash plug-in being activated after waking from sleep.
Comment 12 Anders Carlsson 2015-03-09 13:20:41 PDT
I don't really know if there's anything actionable here. We keep the flash plug-in running for a while even if there are active plug-ins, just to make instantiating the next plug-in faster.
Comment 13 Jim Oase 2015-03-09 13:37:05 PDT
(In reply to comment #12)
> I don't really know if there's anything actionable here. We keep the flash
> plug-in running for a while even if there are active plug-ins, just to make
> instantiating the next plug-in faster.

Good idea if all the computers have wall plugs, questionable idea for those with portables, like my MacBook Pro, and concerns about battery life.

The recent epidemic of MacBook Pro GPU failures highlights the problem with keeping the plug-in active.  Machines with suspect discrete GPUs can run on the integrated GPU. Therefore they can run Safari but not WebKit if they avoid site that use Flash.

The Flash plug-in will activate and stay on after reviving from sleep even when no Flash site has been visited.