1808992 - Motif Window Manager: Maximize button gets Firefox windows extended over the screen's resolution on Linux

Status: UNCONFIRMED

(Core :: Widget: Gtk, defect)

Description

[Sashka [:nochkin]]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/105.0.0.0 Safari/537.36

Steps to reproduce:

1. Linux distribution using MWM (Motif Windows Manager)
2. Install Firefox 73 or higher on Linux
3. Click "Maximize" button (or Alt+F10)

I tried Motif 2.3.4 and Motif 2.3.8 (the latest) on RedHat and Debian.

Actual results:

The Firefox window extends over the screen's working area.
For example, my monitor's resolution is 1920x1200, but maximized Firefox window gets resized to 2800x1600.

Expected results:

The Firefox window should be sized to the screen's resolution. If my resolution is set to 1920x1200, I would expect Firefox to maximize to 1920x1200 at most.

Version of Firefox 72.0.2 and earlier work as expected. It seems like Firefox 73.0 is the first version where this issue was introduced. I also tried Firefox 73.0b1 and I see the same issue.

Comment 1

[Sashka [:nochkin]]

Just to add a note:
This happens on a brand new profile. No extensions or additional customizations. Everything is set to default.
I also tried another hardware configuration with a different video driver and the issue still persists, so it does not seem to be related to my local configuration.
Like I mentioned before, the issue was introduced since Firefox 73.0b1.

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

The Bugbug bot thinks this bug should belong to the 'Core::Widget: Win32' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 3

[Emilio Cobos Álvarez (:emilio)]

(In reply to Sashka from comment #1)

Like I mentioned before, the issue was introduced since Firefox 73.0b1.

If you can, maybe try running:

$ pip3 install --user mozregression
$ mozregression --good 72 --bad 73

That should give us a pushlog where the regression was introduced.

Comment 4

[Sashka [:nochkin]]

Attachment: mozregression results

Thank you for your quick response.
I tried your option using good=72, bad=73 and the first try using 73.0a1 gave me

Comment 5

[Sashka [:nochkin]]

Thank you for your quick response.
I just ran the script and it provided me with the following pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=86afe431ec559c111ae1c1f94e4ede25776e9c50&tochange=1fb1ca98f23b29d1a8a287ad2db326c84df04c57

Here are the narrowed down versions:

 5:28.17 INFO: Narrowed nightly regression window from [2019-12-09, 2019-12-11] (2 days) to [2019-12-09, 2019-12-10] (1 days) (~0 steps left)
 5:28.17 INFO: Got as far as we can go bisecting nightlies...
 5:28.17 INFO: Last good revision: 86afe431ec559c111ae1c1f94e4ede25776e9c50 (2019-12-09)
 5:28.17 INFO: First bad revision: 1fb1ca98f23b29d1a8a287ad2db326c84df04c57 (2019-12-10)

Just in case, I'm also attaching the full run report to this bug.

Comment 6

[Sashka [:nochkin]]

Please disregard the comment #4, it has some incorrect information.

Comment 7

[Sashka [:nochkin]]

Attachment: mozregression-72-73.txt

Comment 8

[Emilio Cobos Álvarez (:emilio)]

Bug 1602563 is the only one touching somewhat related code...

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/105.0.0.0 Safari/537.36

Did you file it from Chrome on Windows? Or are you spoofing the user agent somehow? Does this repro in a clean profile? Can you attach your about:support info? Thanks!

Comment 9

[Sashka [:nochkin]]

Yes, I'm filing it using a different browser and a different device (Chrome on Windows). Sorry for confusions.

I did all tests using Firefox with a clean profile to rule out any issues with my profile.
I can do "about:support". Please let me know which version you need. The latest one that works or the earliest one that does not work?

Comment 10

[Sashka [:nochkin]]

I think I found the change which broke that behavior.

374df506a9d412a961df2f3a90e67413434a73f8	Martin Stransky — Bug 1602563 - [Linux] Allow nsWindow::Resize() reset aspect ratio, r=jhorak

It points to this diff:
https://hg.mozilla.org/mozilla-central/rev/374df506a9d412a961df2f3a90e67413434a73f8

Just to quickly test it out, I commented out line 922 and rebuilt it. Now it seems to work as expected using my test environment.

919:    if (mSizeConstraints.mMaxSize !=
920:        LayoutDeviceIntSize(NS_MAXSIZE, NS_MAXSIZE)) {
921:      AddCSDDecorationSize(&geometry.max_width, &geometry.max_height);
922:      // hints |= GDK_HINT_MAX_SIZE;
923:    }

I would assume commenting out that line is not a fix as it may or may not affect something else. But at least it gives an idea what may be the issue.

Comment 11

[Darkspirit]

(In reply to Sashka from comment #9)

I can do "about:support". Please let me know which version you need. The latest one that works or the earliest one that does not work?

From
$ mozregression --launch 2023-01-09 -a about:support

Comment 12

[Sashka [:nochkin]]

Attaching the screenshot of "about:support" using Firefox 110.0a1 (2023-01-09)

User-agent is:
Mozilla/5.0 (X11; Linux x86_64; rv:109.0) Gecko/20100101 Firefox/110.0

Comment 13

[Sashka [:nochkin]]

Attachment: about-support.PNG

Comment 14

[Darkspirit]

Please click on "Copy text to clipboard" and paste it here. Thanks!

Comment 15

[Sashka [:nochkin]]

Attachment: firefox-about-support.txt

Comment 16

[Sashka [:nochkin]]

Hello,
Just wanted to check if you need any additional information from me on this.
Also, to see if you had a chance to see my message #10 where I think I found the actual change which caused it, and the work-around I did to make it work just to see if this is the source.