1587155 - Color fringing/image corruption on Windows provided driver 22.19.162.4 for Radeon RX 560

Status: RESOLVED FIXED

(Core :: Graphics: WebRender, defect, P2)

Description

[Asif Youssuff]

Attachment: JA7oqKt.png

Reported on reddit: https://www.reddit.com/r/firefox/comments/ddidht/firefox_greenpink_textimages_glitch_how_do_i_fix/

The driver seems to be the latest provided by Windows Update. Updating the driver to the latest provided by the vendor seems to resolve the issue.

Comment 1

[Asif Youssuff]

Attachment: 2cZO5TG.png

Comment 2

[Asif Youssuff]

Attachment: 0pYG0z3.png

Comment 3

[Asif Youssuff]

Attachment: about:support

Comment 4

[Asif Youssuff]

[Tracking Requested - why for this release]:

WebRender on AMD is enabled by default now in release, Windows Update driver is affected.

Comment 5

[Jeff Muizelaar [:jrmuizel]]

We have RX 460 that we'll try to reproduce it on.

Comment 6

[Aria Beingessner [:Gankra]]

This looks like the return of Bug 1565663, which we also failed to reproduce with the RX 460.

Last time we closed it after the user updated their ~2017 driver and it went away. It seems the exact same thing happened here.

Also interesting is that in the old version, it was actually happening regardless of Webrender (for some users?).

Comment 7

[Aria Beingessner [:Gankra]]

In that previous report it was also Driver 22.19.162.4, which suggests we should maybe blacklist this exact driver?

Comment 8

[Aria Beingessner [:Gankra]]

Hey darkspirit, you were able to reproduce some of the similar issues reported, right? Or were those just A Different But Similar Bug?

Comment 9

[Ryan VanderMeulen [:RyanVM]]

Too late to do anything about it in 69 at this point.

Comment 10

[Liz Henry (:lizzard) (relman/hg->git project)]

Likely too late for 70 as well unless we can figure out specific drivers to block.
Jessie, can you help find someone to investigate?

Comment 11

[Jessie [:jbonisteel] pls NI]

Alexis, can you write a patch to block 22.19.162.4? Sounds like at least a good interim solution

Comment 12

[Aria Beingessner [:Gankra]]

Attachment: Bug 1587155 - Block all hardware acceleration for AMD driver 22.19.162.4.

Driver's Haunted.

Comment 13

[Aria Beingessner [:Gankra]]

PSA: I've never blocklisted a driver before, so not 100% sure if this is right, but this is the basic idea at least.

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/f2502597e034
Block all hardware acceleration for AMD driver 22.19.162.4. r=jrmuizel

Comment 15

[Dorel Luca [:dluca]]

https://hg.mozilla.org/mozilla-central/rev/f2502597e034

Comment 16

[Aria Beingessner [:Gankra]]

So we want to uplift this to 70?

Comment 17

[Liz Henry (:lizzard) (relman/hg->git project)]

I think so. Can you request the uplift?

Comment 18

[Aria Beingessner [:Gankra]]

Comment on attachment 9100291 [details]
Bug 1587155 - Block all hardware acceleration for AMD driver 22.19.162.4.

Beta/Release Uplift Approval Request

• User impact if declined: Some users are seeing random nasty visual corruptions on this driver. It goes away if they manually request a driver update, but for whatever reason windows won't automatically do it. We couldn't locally reproduce the issues, so this is the best we can do at the moment.
• Is this code covered by automated tests?: No
• Has the fix been verified in Nightly?: No
• Needs manual test from QE?: No
• If yes, steps to reproduce:
• List of other uplifts needed: None
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): It blocks a single driver version from using hardware acceleration, so exactly those users will just get a slower version of firefox.
• String changes made/needed:

Comment 19

[Liz Henry (:lizzard) (relman/hg->git project)]

Comment on attachment 9100291 [details]
Bug 1587155 - Block all hardware acceleration for AMD driver 22.19.162.4.

Blocks a particular driver, OK for uplift for the RC build.

Comment 20

[Noemi Erli[:noemi_erli]]

https://hg.mozilla.org/releases/mozilla-beta/rev/b6d82b1a6b02

Comment 21

[Darkspirit]

(In reply to Alexis Beingessner [:Gankra] from comment #8)

Hey darkspirit, you were able to reproduce some of the similar issues reported, right? Or were those just A Different But Similar Bug?

(Sorry!) Yes, everything I knew is in bug 1565663. (already in "see also")