Open Bug 1672489 Opened 4 years ago Updated 2 years ago

Don't restrict DirectComposition to Win10 on Nightly.

Categories

(Core :: Graphics: WebRender, task)

task

Tracking

()

REOPENED

People

(Reporter: jrmuizel, Assigned: jrmuizel)

References

(Blocks 1 open bug, Regressed 1 open bug)

Details

Attachments

(5 files, 1 obsolete file)

No description provided.
Pushed by jmuizelaar@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/3714bd86365f Don't restrict DirectComposition to Win10 on Nightly. r=aosmond
Blocks: wr-win8
Type: defect → task
Status: ASSIGNED → RESOLVED
Closed: 4 years ago
Resolution: --- → FIXED
Target Milestone: --- → 84 Branch

Hey guys, I just stumbled across an issue on Windows 7 with 83.0b4 where the share screen on https://janus.conf.meetecho.com/screensharingtest.html will remain black, where as on 84.0a1 (2020-10-25) seems to be working. I searched for a fix, and mozregression got me to this respective issue.

Jeff, can you eventually ask for an uplift for 83 as well?

Flags: needinfo?(jmuizelaar)

(In reply to Catalin Sasca, QA [:csasca] from comment #4)

Hey guys, I just stumbled across an issue on Windows 7 with 83.0b4 where the share screen on https://janus.conf.meetecho.com/screensharingtest.html will remain black, where as on 84.0a1 (2020-10-25) seems to be working. I searched for a fix, and mozregression got me to this respective issue.

That seems wrong. Did this change accidentally disable WebRender for you by chance?

Flags: needinfo?(jmuizelaar) → needinfo?(catalin.sasca)

Hmm, it seems like so. On 83.0b4 it has webrender, and on 84.0a1 it works with Direct3D 11. So it seems that with webrender I will get a black screen, and with Direct3D screenshare will work fine. The gpu is an Intel mobile HD 3000.

Flags: needinfo?(catalin.sasca)
Backout by malexandru@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/56468dae716a Backed out changeset 3714bd86365f for accidentally disabling wr on Win7.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Target Milestone: 84 Branch → ---
Regressions: 1673572

There's a r+ patch which didn't land and no activity in this bug for 2 weeks.
:jrmuizel, could you have a look please?
For more information, please visit auto_nag documentation.

Flags: needinfo?(jmuizelaar)

We shipped DC.

Status: REOPENED → RESOLVED
Closed: 4 years ago2 years ago
Flags: needinfo?(jmuizelaar)
Resolution: --- → FIXED

Could this bug be reopened, please? I've just checked this on my Windows 8.1 PC, and it seems that DirectComposition is still limited to Windows 10 and above on the latest Firefox Nightly 105, whereas this restriction was supposed to be lowered to ≥ Win8.

Flags: needinfo?(gwatson)

Jeff, thoughts on enabling this for Win8.1?

Status: RESOLVED → REOPENED
Flags: needinfo?(gwatson) → needinfo?(jmuizelaar)
Resolution: FIXED → ---

I talked to others on the gfx team, and we're a little concerned at enabling DC by default on Win8.1 due to possible driver quality issues.

I believe you should be able to opt-in to it by going to about:config and setting both gfx.webrender.compositor and gfx.webrender.compositor.force-enabled to true, and restarting the browser.

If you do that, would you be able to post the output of your about:support here and a brief report of whether it's working OK for you?

Flags: needinfo?(jmuizelaar) → needinfo?(kanapa1)
Attached file about:config — desktop (obsolete) (deleted) —
Attached file about:support — desktop (deleted) —
Attachment #9290463 - Attachment is obsolete: true
Attached file about:support — laptop (deleted) —

For a few days now, I have been using the latest Firefox stable with both gfx.webrender.compositor and gfx.webrender.compositor.force-enabled set to "true" (the former was already enabled by default) on my two devices running Windows 8.1. I should have probably used Firefox Nightly instead, but since Firefox stable is my web browser of choice that I use on a daily basis, it was just more convenient for me to test it there.

Anyway, during normal Internet browsing on both devices, I have not encountered any driver or WebRender crashes. Which may or may not be a bit of a surprise, considering that the drivers used are a bit old, especially the one in the laptop. And yet they work just fine. I have not noticed any graphical glitches either, although I may not be observant enough.

However, on the other hand, I have not observed any performance or smoothness improvements or anything like that after setting gfx.webrender.compositor.force-enabled to "true". If there is something I could test in particular, please let me know.

I've attached the about:support content (from both devices) you asked for.

Flags: needinfo?(kanapa1)

If there is any additional information that may (or may not) help push this bug forward, please let me know.

Flags: needinfo?(gwatson)

Could you post the about:support in text format (not like the json above) for your machine with it enabled?

Even if it is enabled, I wouldn't necessarily expect any obvious performance improvements as the resolutions those machines are running at is quite low, so in general the GPU should be good enough to do the compositing while maintaining frame rate. What you might notice if DC is enabled and working well though is lower power usage / better battery life, if those machines are laptops.

Flags: needinfo?(gwatson) → needinfo?(kanapa1)

Thank you very much for your quick reply. When I first attached the data from about:support when you asked me to do so, I chose the raw JSON variant, as it seemed to me that it contained more data that could potentially be important. Additionally, unlike the plain text data, JSONs are not localized.

Anyway, I will attach the new data from the current version of Firefox stable in just a moment.

Flags: needinfo?(kanapa1)
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: