1814482 - White boarders around the window after minimizing

Status: UNCONFIRMED

(Core :: Graphics, defect, P3)

Description

[Wingless1992]

Attachment: IMG_5545.jpg

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/109.0

Steps to reproduce:

Minimize all firefox browser windows

Actual results:

After maximizing windows there is a white border around the two sides and bottom

Expected results:

No borders.

Comment 1

[Wingless1992]

This happens on my Samsung 49in monitor. The only way i've found to correct is to either go full screen. Either by double clicking in the "Tab" bar or hitting F11.

Comment 2

[Rares Doghi, Desktop QA]

I think this issue is a Dupe for Bug 1528747, but I'm not entirely sure, I'm changing the component to Core Graphics for it and maybe one of our devs can take a look.

Comment 3

[Wingless1992]

(In reply to Rares Doghi from comment #2)

I think this issue is a Dupe for Bug 1528747, but I'm not entirely sure, I'm changing the component to Core Graphics for it and maybe one of our devs can take a look.

Sounds good. It's really weird that it does it, lol. Super annoying too. Hopefully they can get it resolved.

Comment 4

[Wingless1992]

Seems to be an issue when two monitors hooked up, that is when i'm experiencing the problem. With only 1 display. I can't reproduce it.

Comment 5

[Jamie Nicol [:jnicol]]

Hi Wingless, thank you for the bug report.

Could you please navigate to about:support, click "copy text to clipboard", and attach the contents of the clipboard to this bug.

Do you know whether this used to work and the bug has at some point been introduced, or has this always been the case?

Comment 6

[Wingless1992]

Attachment: Clipboard attachment

Comment 7

[Wingless1992]

(In reply to Jamie Nicol [:jnicol] from comment #5)

Hi Wingless, thank you for the bug report.

Could you please navigate to about:support, click "copy text to clipboard", and attach the contents of the clipboard to this bug.

Do you know whether this used to work and the bug has at some point been introduced, or has this always been the case?

Not sure. I used to use Firefox years ago but switched to Chrome. I'm now back after a long hiatus.

Comment 8

[Jamie Nicol [:jnicol]]

Thanks. Could you try going to about:config, setting gfx.webrender.compositor to false, restarting firefox, and see if you can still reproduce the bug?

Comment 9

[Wingless1992]

(In reply to Jamie Nicol [:jnicol] from comment #8)

Thanks. Could you try going to about:config, setting gfx.webrender.compositor to false, restarting firefox, and see if you can still reproduce the bug?

Made the change, still does it. Hmmmm.

Comment 10

[BugBot [:suhaib / :marco]]

The severity field is not set for this bug.
:bhood, could you have a look please?

For more information, please visit auto_nag documentation.

Comment 11

[Bob Hood [:bhood]]

Slim chance, but could you update your NVIDIA driver from 528.24 to the current (528.49) and see if that has any effect?

Comment 12

[Wingless1992]

(In reply to Bob Hood [:bhood] from comment #11)

Slim chance, but could you update your NVIDIA driver from 528.24 to the current (528.49) and see if that has any effect?

Nope, still the same problem. It'll be interesting to see what's actually causing this as it's super annoying.

Comment 13

[Bob Hood [:bhood]]

Glenn, need your wisdom here. Any insights you can provide?

Comment 14

[Glenn Watson [:gw]]

It might be related to the high resolution and mixed display scale ratios / refresh rates. Just speculating as that's an uncommon configuration:

Display0: 5120x1440@240Hz scales:1.000000|1.000000
Display1: 3840x2160@60Hz scales:1.000000|2.000000

Could you try physically unplugging one monitor, seeing if you can reproduce with a single monitor plugged in, and then repeat the experiment with the other monitor?

Comment 15

[Glenn Watson [:gw]]

This line is also interesting from about:support:

HardwareStretching: both=0 window-only=0 full-screen-only=2 none=0 error=0

That suggests that the driver is reporting that DWM can only do hardware stretching in full-screen, if I'm reading correctly, which seems unexpected for that GPU?

Comment 16

[Wingless1992]

Did some testing. It only happens when both monitors are in use. Only use 1 and it's fine, even at higher refresh rates. Testing just refresh rates on my main 49in monitor it doesn't do anything, works fine. For some weird reason it's only when both monitors are in use. Chrome doesn't have this problem so I don't think its a driver issue with Nvidia or a hardware problem.

This setup is kind of unusual that's for sure. I use a Samsung 49in 240hz monitor as my main and then a Vizio 4k 60hz TV has a 2nd monitor. The TV is there just for background video and occasional sports watching. Even playing with the refresh rates doesn't help. I set both to 60hz and it still does it.

For now, i've unplugged the 2nd monitor and it seems to be working as it should. I wonder if I changed something I shouldn't have in the about:config. Would uninstalling firefox, redownloading it and installing it change anything? Meaning would it reset the about:config?

Comment 17

[Ashley Hale [:ahale]]

When I've experimented with triggering device resets (by overloading the card and exhausting vram) with mixed refresh rates I got the browser in such a state a few times but I didn't try to characterize the bug yet, it was kind of rare but predictable enough that I triggered it multiple times. I can take a look.

It's my vague understanding that the GPU actually has a global limited number of hardware scanout planes (MPO) that are spread across the number of display port channels going to monitors, so if you have a 5120@240hz monitor it can reduce the number of overlays on all monitors because it takes 4 channels to drive that one monitor, leaving fewer channels for the others and the hardware overlays compete for that same limit.

But the white border seems more related to the mixed refresh workaround and seemed stateful - triggering device resets while the mixed refresh workaround is triggering caused a different swapchain size of the window for some reason.

Comment 18

[Glenn Watson [:gw]]

OK, that sounds like we're starting to identify what may trigger the bug, at least. One more experiment - what if you have both monitors connected, but the scaling ratio is the same on both monitors (i.e. both monitors at 1x scale, or both monitors at 2x scale, rather than the default mixed scale you have)?

Comment 19

[Glenn Watson [:gw]]

In answer to your question - it's possibly worth trying with a new / fresh profile (run from the command line with -p) which might be easier to uninstalling, re-downloading etc, but I think it's unlikely this will make a difference. I suspect it's a bug, perhaps related to how we handle window sizing in multi-monitor set ups with mixed scaling ratios, perhaps something weird we're triggering in the driver / dwm.

Comment 20

[Wingless1992]

(In reply to Glenn Watson [:gw] from comment #18)

OK, that sounds like we're starting to identify what may trigger the bug, at least. One more experiment - what if you have both monitors connected, but the scaling ratio is the same on both monitors (i.e. both monitors at 1x scale, or both monitors at 2x scale, rather than the default mixed scale you have)?

I set the scaling of the 2nd monitor to 100% instead of the 200% I had it before. It seems like it fixed it but i'm not 100% sure. Windows recommends 300% scaling i'm guessing because of the differences in monitor resolution.

Either way. Lemme play with it for awhile and see if its either fixed or the issue is still present. Thanks y'all!

Comment 21

[Wingless1992]

I just did my end of day routine. Minimize all windows. Shut the 2nd monitor off, then the main. Waited about 1min then turned everything back on in reverse order.

NO WHITE BARS!!! Lol.

Again, i'll keep testing but if this solves it, hell yea.

Comment 22

[Glenn Watson [:gw]]

That's great if there's a work around, we should aim to fix the underlying bug though. Not sure if it's more likely to be a gfx or widget component bug. Sotaro, ideas where / who might be best to investigate this?

Comment 23

[Wingless1992]

(In reply to Glenn Watson [:gw] from comment #22)

That's great if there's a work around, we should aim to fix the underlying bug though. Not sure if it's more likely to be a gfx or widget component bug. Sotaro, ideas where / who might be best to investigate this?

For sure, just let me know what you need/want me to do. More than happy to help the community out.