User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:84.0) Gecko/20100101 Firefox/84.0

Steps to reproduce:

Play a TV-livestream of German TV

Actual results:

CPU usage is near 50% of a 4 core CPU, not constant as usual.

Expected results:

Should be only ca 25% CPU usage as it was before (V <= 83)
Also VLC streams in the same environment and the same URL with ca. 25% CPU usage, as it always was.

Bugbug thinks this bug should belong to this component, but please revert this change in case of error.

Hi, unfortunately this report is not very useful because it does not describe the problem well. If you have time and can still reproduce the situation or problem, please read https://developer.mozilla.org/en-US/docs/Mozilla/QA/Bug_writing_guidelines and add a more useful description to this report, with clear and exact steps to reproduce.

I will do my very best, to bring up needed info. But I'm not so familiar with this.
My first attempt is to attach a about:support file (raw file).(Some settings are experimental and maybe useless or wrong.)
Second attempt is a permalink from profiler.firefox.com: : https://share.firefox.dev/3kUnJPI - maybe this could be usual for you.
So I need some help, if this is not enough or unexperienced.

Could you try navigating to about:config and setting the gfx.webrender.force-disabled pref to true, then restarting Firefox and seeing if that has any impact on the issue? You can set the pref back after testing.

Beside your suggestion I tested some more configurations of webrender in about:config, but I could not find a setting, which reduces the CPU usage.

It seems to be possible, that this is a kde plasma issue. One Test with ICE-WM showed me a normal CPU usage. There is a [Wayland] KDE bug tracker Bug 1609115, maybe this bug belongs there too.

In comment 0 it's mentioned that this wasn't an issue in version 83. Could you test the version again to confirm that is the case? It would be useful to know if this is something that regressed in Firefox. You could use mozregression to find the specific version, which would be very helpful.

This is the result of
MOZ_ENABLE_WAYLAND=1 ~/.local/bin/mozregression --good 2020-11-01 --bad 2020-11-17 --pref gfx.webrender.all:true layers.acceleration.force-enabled:true

20:52.65 INFO: No more integration revisions, bisection finished.
20:52.65 INFO: Last good revision: 0d82c3f140ba2110d3d40512c9f04b4cadcb4883
20:52.65 INFO: First bad revision: 67c0ee5d0b8567fe842bf3e3e1c757372e24dfaa
20:52.65 INFO: Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=0d82c3f140ba2110d3d40512c9f04b4cadcb4883&tochange=67c0ee5d0b8567fe842bf3e3e1c757372e24dfaa

Sometimes it was not quite easy to decide, because of the download in the background.

This looks like issue with dmabuf wayland backend. Please try to disable:

media.ffmpeg.dmabuf-textures.enabled
media.ffmpeg.vaapi.enabled
widget.wayland-dmabuf-vaapi.enabled
widget.wayland-dmabuf-video-textures.enabled

Can you attach a log of the playback?

(In reply to Martin Stránský [:stransky] from comment #10)

This looks like issue with dmabuf wayland backend. Please try to disable:

media.ffmpeg.dmabuf-textures.enabled
media.ffmpeg.vaapi.enabled
widget.wayland-dmabuf-vaapi.enabled
widget.wayland-dmabuf-video-textures.enabled

These are my settings in about:config (copied from there). Search-items are "vaapi", "dma", "widget.waylad":

media.ffmpeg.vaapi-drm-display.enabled true
media.ffmpeg.vaapi.enabled false
vaapi
BooleanNumberString

app.update.backgroundMaxErrors 10
browser.newtabpage.activity-stream.logowordmark.alwaysVisible false
media.ffmpeg.dmabuf-textures.disabled true
widget.dmabuf-webgl.enabled true
dma
BooleanNumberString

widget.wayland-smooth-rendering false
widget.wayland.use-opaque-region false
widget.wayland_vsync.enabled false
widget.wayland
BooleanNumberString

Could not find:

widget.wayland-dmabuf-vaapi.enabled
widget.wayland-dmabuf-video-textures.enabled
It is FF 85.0b5.

(In reply to Martin Stránský [:stransky] from comment #11)

Can you attach a log of the playback?

With this line I got only an empty log file:
firefox --allow-downgrade --MOZ_LOG_FILE=/home/walther/Änderungen_Anleitungen/bugs/firefox/Bug_1678806/firefox.log
What is wrong?
New attachment: I added a file from walther@localhost:~> firefox --allow-downgrade --MOZ_LOG_FILE=/home/walther/Änderungen_Anleitungen/bugs/firefox/Bug_1678806/firefox.log as "bash.txt"

Testing FF-nightly (appVersion="86.0a1" buildID="20210107040715) with gfx.webrender.software=enabled. CPU-usage is still significant higher than with FF 85.0b5 !

(In reply to W.Pelser from comment #15)

Testing FF-nightly (appVersion="86.0a1" buildID="20210107040715) with gfx.webrender.software=enabled. CPU-usage is still significant higher than with FF 85.0b5 !

This is stdout of appVersion="86.0a1" buildID="20210111093448" with gfx.webrender.software=enabled:
@localhost:~> /home//.firefox/nightly/firefox/firefox https://www1.wdr.de/mediathek/video/live/videostream-wdr-fernsehen-im-livestream--102.html
Sandbox: seccomp sandbox violation: pid 9676, tid 9685, syscall 203, args 9685 128 140210563960112 8 140210563962432 1. Killing process.
Sandbox: seccomp sandbox violation: pid 9676, tid 9684, syscall 203, args 9684 128 140210572352816 8 140210572355136 1. Killing process.
Sandbox: crash reporter is disabled (or failed); trying stack trace:
Sandbox: frame #01: pthread_setaffinity_np[/lib64/libpthread.so.0 +0x14481]
Sandbox: frame #02: ???[/usr/lib64/dri/i965_dri.so +0x556704]
Sandbox: frame #03: ???[/usr/lib64/dri/i965_dri.so +0x5559d7]
Sandbox: frame #04: ???[/lib64/libpthread.so.0 +0x93e9]
Sandbox: frame #05: clone[/lib64/libc.so.6 +0x1019c3]
Sandbox: frame #06: ??? (???:???)
Sandbox: end of stack.

Should I open a new bug with this?

Do you still see that with latest nightly?
If so please run Firefox on terminal with MOZ_LOG="PlatformDecoderModule:5" env variable and attach the log here.
Thanks.

Please reopen if you have the info.
Thanks.