1991771 - Sluggish performance and all-black windows

Status: RESOLVED WORKSFORME

(Thunderbird :: General, defect)

Description

[Tristan Miller]

Attachment: Terminal output from Thunderbird

I'm running openSUSE Tumbleweed on a ThinkPad T14 with an AMD Radeon 780M (RADV PHOENIX). Following a system update yesterday, Thunderbird 143.0.1 (both the official binaries and the openSUSE RPMs) stops working properly after a few minutes' use: the main and composer windows become entirely black and stop repainting properly. It's still possible to type and blindly click around, but the responsiveness is very sluggish. Meanwhile lots of error messages are printed to the terminal:

ATTENTION: default value of option mesa_glthread overridden by environment.
[GFX1-]: Detect DeviceReset DeviceResetReason::UNKNOWN DeviceResetDetectPlace::WR_POST_UPDATE in Parent process

ATTENTION: default value of option mesa_glthread overridden by environment.
[GFX1-]: Failed to compile vertex shader: cs_clip_rectangle

[ERROR webrender::device::gl] Failed to compile vertex shader: cs_clip_rectangle
    
[GFX1-]: Failed to compile vertex shader: cs_blur_COLOR_TARGET

[ERROR webrender::device::gl] Failed to compile vertex shader: cs_blur_COLOR_TARGET
    
[GFX1-]: Failed to compile vertex shader: cs_border_solid

See attached for more.

System updates included going from kernel-default 6.16.7 to 6.16.8, and maybe also kernel-firmware-amdgpu 20250825 to 20250916.

The symptoms look a lot like what's described in Bug 1986254 and its duplicate Bug 1989579, so perhaps this is a duplicate as well.

Comment 1

[Wayne Mery (:wsmwk)]

(In reply to Tristan Miller from comment #0)

...
System updates included going from kernel-default 6.16.7 to 6.16.8, and maybe also kernel-firmware-amdgpu 20250825 to 20250916.

The symptoms look a lot like what's described in Bug 1986254 and its duplicate Bug 1989579, so perhaps this is a duplicate as well.

Other linux performance reports https://mzl.la/48FJejA
There were both duped to Bug 1667748 - Should we open our files with O_CLOEXEC?

Comment 2

[Wayne Mery (:wsmwk)]

How is 144? Did you look at the other bug reports?

Comment 3

[Tristan Miller]

No longer reproducible for me following an update to 144 (and perhaps also to more recent AMD GPU firmware, Mesa, etc.)