[JAY] Right clicking creates wayland region with negative width or height, causing a crash on jay which rejects this as a protocol error
Categories
(Core :: Widget: Gtk, defect)
Tracking
()
People
(Reporter: svante.b37, Unassigned, NeedInfo)
References
Details
Attachments
(1 file)
|
4.85 MB,
text/x-log
|
Details |
User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:160.0) Gecko/20100101 Firefox/160.0
Steps to reproduce
I run the jay wayland compositor on a display with a scale factor of 1.2, when I right-click in the main browsing area, most often on the right size of my screen, firefox crashes and the following error is logged:
(jay) wl_display#1: error 3: An error occurred while processing a request:
Could not process awl_region#101.addrequest: width and/or height are negative
This has been going on for a few months by now and I'm slowly training myself to never right-click and open links in new tabs when I don't have a middle-click by copying the current URL, opening a new tab, pasting the URL and clicking the link I intended. It's painful, but it works.
Actual results
Firefox attempted to create a wayland region with a negative size, which jay rejected as a fatal error and disconnected Firefox, which then crashed.
I've sent several crash reports about this over the last 3 months, such as these four from today bp-6b95ae28-c02a-4e91-a7bb-fb3e50261106, bp-f3b9dc72-97c7-4e43-a330-8c7d50261106, bp-751ca542-f8c8-409b-828e-e5c510261106, bp-51587e49-e407-4876-ad5a-0bcb8026110
Expected results
The context menu should've shown with no crash
Comment 1•14 days ago
|
||
The Bugbug bot thinks this bug should belong to the 'Core::Widget: Gtk' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.
Comment 2•14 days ago
|
||
Please run on terminal with MOZ_LOG="Widget:5,WidgetPopup:5" WAYLAND_DEBUG=1 env variables and attach the log here.
Thanks.
Updated•13 days ago
|
Uploaded log related to crash bp-0dc414c5-ce6a-4784-8f83-46ef30261106
Oh, and sorry it doesn't happen 100% of the time, and in this one I ended up having to right click a few times. Look around [22:18:54.485232], I believe
Comment 6•10 days ago
|
||
[Parent 480640: Main Thread]: V/WidgetPopup [7f3d9df51000]: nsWindow::FractionalScaleFactor(): fractional scale 1.20
[Parent 480640: Main Thread]: V/WidgetPopup [7f3d9df51000]: nsWindow::FractionalScaleFactor(): fractional scale 1.20
[Parent 480640: Main Thread]: V/WidgetPopup [7f3d9df51000]: nsWindow::FractionalScaleFactor(): fractional scale 1.20
[Parent 480640: Main Thread]: V/WidgetPopup [7f3d9df51000]: nsWindow::FractionalScaleFactor(): fractional scale 1.20
[Parent 480640: Main Thread]: V/WidgetPopup [7f3d9df51000]: nsWindow::FractionalScaleFactor(): fractional scale 1.20
[Parent 480640: Main Thread]: D/WidgetPopup [7f3d9df51000]: nsWindow::UpdateOpaqueRegionInternal() set opaque region [221,10] -> [60 x 0]
[Parent 480640: Main Thread]: V/WidgetPopup [7f3d9df51000]: nsWindow::FractionalScaleFactor(): fractional scale 1.20
[Parent 480640: Main Thread]: D/WidgetPopup [7f3d9df51000]: nsWindow::UpdateOpaqueRegionInternal() set opaque region [5,10] -> [285 x 396]
[22:18:54.485105] {Default Queue} -> wl_compositor#32.create_region(new id wl_region#208)
[22:18:54.485116] {Default Queue} -> wl_region#208.add(5, 10, 285, 396)
[22:18:54.485127] {Default Queue} -> wl_surface#157.set_opaque_region(wl_region#208)
[22:18:54.485130] {Default Queue} -> wl_region#208.destroy()
[22:18:54.485217] {Default Queue} -> wl_compositor#32.create_region(new id wl_region#180)
[22:18:54.485232] {Default Queue} -> wl_region#180.add(221, 10, 60, -1)
[22:18:54.485264] {Default Queue} -> wl_region#180.add(5, 10, 285, 396)
[22:18:54.485307] {Default Queue} -> wl_surface#132.frame(new id wl_callback#121)
[22:18:54.485315] {Default Queue} -> wl_surface#132.commit()
Looks like [Parent 480640: Main Thread]: D/WidgetPopup [7f3d9df51000]: nsWindow::UpdateOpaqueRegionInternal() set opaque region [221,10] -> [60 x 0]
is transferred to -1 so looks like Bug 2073312 is related here.
Comment 7•10 days ago
|
||
Yes, the issue looks valid and it's caused by the round-in. I wonder why we don't see it on Mutter/KDE.
Description
•