Closed Bug 1629041 Opened 4 years ago Closed 3 years ago

[Sway] [Wayland] GLXtest process failed

Categories

(Core :: Widget: Gtk, defect, P3)

77 Branch
defect

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: AlexanderMeissner, Unassigned)

References

(Blocks 1 open bug)

Details

Attachments

(3 files)

Attached file debugging.json

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.13; rv:77.0) Gecko/20100101 Firefox/77.0

Steps to reproduce:

I ran the official firefox 75 release (installed via pacman) on arch linux in sway with the env MOZ_ENABLE_WAYLAND=1 .

Actual results:

Everything webgl related is disabled by FEATURE_FAILURE_GLXTEST_FAILED.
graphics - adapterDescription: GLXtest process failed (exited with status 1): Unable to open a connection to the X server.
graphics - windowProtocol: x11

Expected results:

Clearly the windowProtocol should be wayland as there is no x11 on sway.
Also, it should not try to connect to the X server and use EGL instead of GLX.

Bugbug thinks this bug should belong to this component, but please revert this change in case of error.

Component: Untriaged → Widget: Gtk
Product: Firefox → Core
Blocks: wayland-sway

Resetting severity to default of --.

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

Severity: normal → --

I'm experiencing the same issue. Installed firefox-developer-edition via pacman on Arch Linux also (77.0b9), using sway as window manager.

Attached file firefox-debugging.json

Robert, am I right that you have done some investigation here?

Flags: needinfo?(robert.mader)
Depends on: 1640053

TBH I'm confused - the way this has been implemented in bug 1556301 (1) should make sure that we only fall back to GLX if the DISPLAY env variable is set, implying Xwayland is available, or if the egl test failed. Maybe what's happening here is that something in they Wayland / EGL stack fails and we fall back to GLX, thus printing the wrong/unrelated errors. Given both these errors happened on Arch, one random guess would be that some Mesa library was not properly installed as dependency (e.g. libGL.so.1, libGLESv2.so.2, libEGL.so.1). Greg, have you experienced issues with this lately?

That being said, in bug 1640053 I try to clean up the code a bit more (while trying to push the mesa devs to implement what we need to fully rely on EGL, without X connection on Wayland). And to allow a fallback from GL to GLES, e.i. libGL.so.1 to libGLESv2.so.2.

https://searchfox.org/mozilla-central/source/toolkit/xre/glxtest.cpp#558

Flags: needinfo?(robert.mader) → needinfo?(greg)
Assignee: nobody → robert.mader
Status: UNCONFIRMED → ASSIGNED
Ever confirmed: true
Status: ASSIGNED → NEW
Assignee: robert.mader → nobody

I haven't experienced any problems — but I run without Xwayland, so glxtest can't even try the X test.

we only fall back to GLX if the DISPLAY env variable is set, implying Xwayland is available, or if the egl test failed

No, if DISPLAY is set, implying Xwayland is available, the GLX test is preferred.

This wasn't the case in my initial version, but people were complaining that the info you want to collect in bug 1640053 via new mesa exts broke WebRender qualification or something >_<

Flags: needinfo?(greg)

(the LACK OF that info, lol)

Attached file Firefox 77 debug info
I'm also experiencing this on wayland-sway with firefox 73.0.1 and 79.0

I'm also experiencing this on wayland-sway with firefox 73.0.1 and 79.0

Debug info from about:support:
https://pastebin.com/xumUTaPY

Priority: -- → P3

This one was already fixed in some other bug a while ago, after bug 1640053 it should be even better :)

Status: NEW → RESOLVED
Closed: 3 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: