Closed Bug 1390481 Opened 9 years ago Closed 5 years ago

[e10s][APZ] Flash performance rapid decline on firefox52.2.1 (due to repainting invisible animated gif)

Categories

(Core :: Layout: Images, Video, and HTML Frames, defect, P3)

52 Branch
Unspecified
Linux
defect

Tracking

()

RESOLVED INVALID
Tracking Status
firefox-esr52 --- wontfix
firefox55 --- wontfix
firefox56 --- wontfix
firefox57 --- wontfix
firefox58 --- wontfix
firefox59 --- fix-optional

People

(Reporter: hwjeastd07, Unassigned)

References

Details

(Keywords: multiprocess, perf, Whiteboard: [gfx-noted])

Attachments

(9 files)

User Agent: Mozilla/5.0 (Windows NT 6.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/59.0.3071.86 Safari/537.36 Steps to reproduce: 1:open link http://tv.sohu.com/20141208/n406764549.shtml, 2:right click on video, select video information to see fsp Actual results: 3: enable e10s, result: fps 7fps. 4: disable e10s, result: fps 25fps Expected results: 5: enable e10s, result: fps 25fps. same as disable e10s
Component: Untriaged → Plug-ins
Product: Firefox → Core
I test again,the fps decline is not associated with e10s; but with apz mechanism (layers.async-pan-zoom.enabled) enable apz : fps 7fps; disable apz : fps 25fps
Summary: [e10s] with E10s enabled, Flash performance rapid decline on firefox52.2.1 → [e10s][APZ] with APZ enabled, Flash performance rapid decline on firefox52.2.1
Flags: needinfo?(jmathies)
Component: Plug-ins → Panning and Zooming
Flags: needinfo?(jmathies)
Keywords: flashplayer
OS: Unspecified → Windows
Please post your about:support text, thanks!
Flags: needinfo?(hwjeastd07)
Flags: needinfo?(hwjeastd07)
Summary: [e10s][APZ] with APZ enabled, Flash performance rapid decline on firefox52.2.1 → [e10s][APZ] linux platform, with APZ enabled, Flash performance rapid decline on firefox52.2.1
Attached image about_support_1.jpg —
Attached image about_support_2.jpg —
Attached image about_support_3.jpg —
See Also: → 1211565
Can you check to see if this happens in Nightly (Firefox 57) as well? 52 ESR will not be getting performance fixes anyway (ESR branches only get security fixes), so unless it affects Nightly we won't investigate this.
Priority: -- → P3
Whiteboard: [gfx-noted]
ok, i try it. it same as bug1211565. Bug 1211565 ,do you have Plan to deal with?
Enable apz, when i play video on sohu. it always trigger UpdateHitTestingTree. but when i play video on other website,such as youku (flash), this phenomenon does not exist. Enable apz, sohu.com 7fps; youku(flash) 25fps; I don't know why sohu always trigger UpdateHitTestingTree. could you give me some suggestions?
UpdateHitTestingTree should only get called if there are changes to the layer tree structure, which is usually caused by some sort of change to the page. If it's just a video playing there should be no change to the layer tree structure. So most likely some javascript is running on the page and changing things as the video is playing.
when i enable apz, the profiler tool display frequently paint. but when i disable apz, the phenomenon does not exist
when i disable profiler, profiler tool display the frequently paint phenomenon doesn't exist.
The phenomenon difference between enable apz and disable apz, exist on firefox52 and inbound branch.
after i test, i found that the gif dynamic background cause the page paint frequently. li tag use this gif as background.
after i delete the gif . start firefox with apz enable, the flash frame rate normal。
Ah, nice work! Presumably we should not be spending time repainting this gif if it's not visible. It might be that this only started happening with APZ because the gif is not immediately visible but is in the displayport which is larger with APZ.
Component: Panning and Zooming → Layout: Images
OS: Windows → Linux
Summary: [e10s][APZ] linux platform, with APZ enabled, Flash performance rapid decline on firefox52.2.1 → [e10s][APZ] Flash performance rapid decline on firefox52.2.1 (due to repainting invisible animated gif)
Product: Core → Core Graveyard
Product: Core Graveyard → Core

We're in the process of removing support for plugins (bug 1677160) and bug 1687239 has removed the relevant Layout code, so I think this bug is irrelevant now.

Status: UNCONFIRMED → RESOLVED
Closed: 5 years ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: