1626959 - WebRender causes Firefox to become insensitive to mouse clicks on RTX 2060

Status: RESOLVED INACTIVE

(Core :: Graphics: WebRender, defect, P3)

Description

[Asif Youssuff]

Originally posted: https://www.reddit.com/r/firefox/comments/ftf795/help_new_pc_with_firefox_wont_respond_to_mouse/

When I launch Firefox I can't click on anything but with the address bar in focus I can type. If I disable hardware acceleration, it works. Launching Firefox in safe mode also works.

Disabling WebRender fixes the issue.

Comment 1

[Asif Youssuff]

Attachment: about:support with WebRender enabled

Comment 2

[Asif Youssuff]

Attachment: about:support with WebRender disabled

Comment 3

[Darkspirit]

I also tried Firefox ESR, same issue.

Same Nvidia driver as bug 1625857:

Driver Version: 26.21.14.4575
Driver Date: 3-17-2020

Comment 4

[Jessie [:jbonisteel] pls NI]

We suspect this is because of a bad driver. Asif, can you recommend that people rollback to the previous driver? It is: 442.74

Comment 5

[Dzmitry Malyshau [:kvark]]

setting P1 matching the wr-76 blocker tag

Comment 6

[Asif Youssuff]

Attachment: about:support

Jessie, the user went back to 442.74 and they see the same issue. `about:support` attached.

Comment 7

[Jessie [:jbonisteel] pls NI]

In the first about:support in comment 1 it looks like the person was using beta, with WR and not DirectComp

In the most recent about:support it looks like they are using release, with WR + DirectComp. Given the issue seems to happen in both cases, makes me think DirectComp is probably not at fault.

In their Reddit comment, they mention using Windows LTSC. Apparently LTSC is:

"The Long-Term Servicing Channel (LTSC) is designed for Windows 10 devices and use cases where the key requirement is that functionality and features don’t change over time. Examples include medical systems (such as those used for MRI and CAT scans), industrial process controllers, and air traffic control devices. These devices share characteristics of embedded systems: they are typically designed for a specific purpose and are developed, tested, and certified before use."

This leads me to believe that it might be something funky with using LTSC.

Asif, if you are able to get in touch with this user again it would be interesting to understand why they are using LTSC especially if they are just on a personal computer.

Comment 8

[Jessie [:jbonisteel] pls NI]

We are going to block Windows LTSC users from getting WebRender, for now: https://bugzilla.mozilla.org/show_bug.cgi?id=1632178

Comment 9

[Asif Youssuff]

I tried LTSC because I read in reddit that it's a leaner version of Windows with no updates being forced on you. Unfortunately LTSC doesn't make games run physx on the GPU so I went back to 1909.

Hi! I'm on Windows 1909 with all the updates now, LTSC was causing other issues with my GPU. But still the browser issue is here on 1909. :/

Jessie, the user sees the issue even when not using LTSC - personally, I think LTSC is a red herring, as it ought to be just a "ESR" version of Windows.

Any further ideas on narrowing down what this might be? The user is back on 1909.

Comment 10

[Jessie [:jbonisteel] pls NI]

Hm, that is really strange. We have a team member with this same card and they do not experience this issue. There is likely something particular to their set up. Maybe a mouse driver issue? Strange though. Do they have the same issue on Nightly?

Comment 11

[Firefox Bug Husbandry Bot]

Because this bug's Severity has not been changed from the default since it was filed, and it's Priority is P3 (Backlog,) indicating it has been triaged, the bug's Severity is being updated to S3 (normal.)

Comment 12

[Asif Youssuff]

User does not see the issue in Nightly, at least not 13 days ago. They continue to see it in dev edition.

Comment 13

[Timothy Nikkel (:tnikkel)]

Random guess, what if they flip apz.windows.force_disable_direct_manipulation to false?

Comment 14

[Asif Youssuff]

(In reply to Timothy Nikkel (:tnikkel) from comment #13)

Random guess, what if they flip apz.windows.force_disable_direct_manipulation to false?

User says that it works if they set apz.windows.force_disable_direct_manipulation to false.

Comment 15

[Timothy Nikkel (:tnikkel)]

Hmm, so from reading the thread the user also saw the issue on ESR

https://www.reddit.com/r/firefox/comments/ftf795/help_new_pc_with_firefox_wont_respond_to_mouse/fm6rue0/

at the time of that comment ESR was 68. ESR 68 never had the window styles that the pref apz.windows.force_disable_direct_manipulation disables. So the ESR they tried should match apz.windows.force_disable_direct_manipulation = false.

So I'm not sure if that pref has anything to do with it, there might be complicating factors going on?