GNOME Bugzilla – Bug 766146
glyph cache contains stale textures after suspend/resume cycle in certain (hardware/software?) configurations
Last modified: 2021-07-05 13:50:28 UTC
Created attachment 327477 [details] Screenshot of missing glyphs in gnome-shell On my system, gnome-shell sometimes does not render certain glyphs. Visually, that looks like the attachment (a screenshot I got from another user who suffers from the same bug). Note how some glyphs are consistently rendered (F, x, W, b) whereas others are missing. Reading the source, I stumbled upon cogl-pango having a "glyph cache", which seemed like a likely place for this malfunction. And indeed, a call to clear the cache using gdb fixes the issue: sudo gdb -p `pgrep -u $USER -x gnome-shell` <<<"print /r _cogl_pango_renderer_clear_glyph_cache((CoglPangoRenderer*)(_cogl_pango_font_map_get_priv(clutter_context_get_pango_fontmap())->renderer))" (this needs the debugging symbols installed). I don't know what, exactly, causes these stale textures, but it seems to happen every now and then after a suspend/resume cycle. It may therefore also be a graphics driver bug. I'm still reporting it here against cogl-pango because this is as far as I've been able to debug the issue with my limited knowledge of the full stack. I observed this issue both under the X session and under Wayland. Here's some version information for my system: Ubuntu 16.04 + gnome-staging ppa (https://launchpad.net/~gnome3-team/+archive/ubuntu/gnome3-staging) ii gnome-shell 3.20.1-0ubuntu1~xenial3 amd64 graphical shell for the GNOME desktop ii libclutter-1.0-0:amd64 1.26.0-2~xenial0 amd64 Open GL based interactive canvas library ii libcogl-pango20:amd64 1.22.0-2 amd64 Object oriented GL/GLES Abstraction/Utility Layer ii libdrm-intel1:amd64 2.4.67-1 amd64 Userspace interface to intel-specific kernel DRM services -- runtime Intel HD Graphics 4000 For reference, here's a question about this on askubuntu, together with my answer detailing, in essence, the same as what I wrote in this bug report: https://askubuntu.com/questions/756275/gnome-shell-text-getting-messed-up/769407 Looking forward to any insight you may have!
*** Bug 762064 has been marked as a duplicate of this bug. ***
*** Bug 766212 has been marked as a duplicate of this bug. ***
*** Bug 766791 has been marked as a duplicate of this bug. ***
*** Bug 766792 has been marked as a duplicate of this bug. ***
Can you all confirm that you're seeing this issue on intel gpus on ubuntu?
I have this bug, and i indeed have an intel gpu, and ubuntu.
Extra info: i have an intel i3 4005u with 4400 graphics
Intel GPU yes (core i5 4300u with HD 4400 graphics), Ubuntu no (using Arch Linux with Gnome-Shell: 3.20.2-1, Kernel: 4.4.11-1-lts, Cogl: 1.22.0+11+g81ab4a3-1, Clutter: 1.26.0-1, libdrm: 2.4.68-1)
Took another screenshot, this time of random text dissapearing. http://imgur.com/3Zzn5xm
*** Bug 767089 has been marked as a duplicate of this bug. ***
I don't know if this is related, because it could just be a formation of this bug https://bugs.launchpad.net/ubuntu/+source/unity/+bug/1532508, but when I opened my laptop lid just now, my screen was completely black for about 10 seconds, then eventually the cursor appeared, then in a while instead of taking me to the locked-screen I saw my Firefox window that I had minimized before closing the laptop lid except for everything above the list of tabs was just black, but the most worrying thing was when I moved my cursor I was actually able to interact with the window and click on things etc, so it actually bypassed the locked-screen, and stayed in this state for at least 10 seconds, it then went black again and finally took me to the locked screen. After logging in I found that Firefox was minimized as it was meant to be, however the bookmark I had clicked on to go to when it had bypassed the locked-screen was open in the tab, so this, whatever it is caused by, is a very dangerous security risk. Do you think that this is a formation of this bug? The other linked bug? Or a separate bug that I should report? The only reason I suspect it could be a formation of this bug is because this bug was getting worse and worse every time with more and more things going wrong and then finally this happens, so it could just be some other bug occurred, a worse version of this, anyway my bug got closed as a duplicate of this one so wasn't sure where else to ask what to do about what I have described above...
I doubt it, they seem like very different issues to me.
You're right, I think I will file a new report then.
Related bug report https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1576246. I am experiencing this issue randomly after waking from sleep on Asus Zenbook UX301LA with Intel Iris™ 5100 graphics.
@timo(In reply to Timo Kluck from comment #0) > Created attachment 327477 [details] > Screenshot of missing glyphs in gnome-shell > > On my system, gnome-shell sometimes does not render certain glyphs. > Visually, that looks like the attachment (a screenshot I got from another > user who suffers from the same bug). Note how some glyphs are consistently > rendered (F, x, W, b) whereas others are missing. > > Reading the source, I stumbled upon cogl-pango having a "glyph cache", which > seemed like a likely place for this malfunction. And indeed, a call to clear > the cache using gdb fixes the issue: > > sudo gdb -p `pgrep -u $USER -x gnome-shell` <<<"print /r > _cogl_pango_renderer_clear_glyph_cache((CoglPangoRenderer*)(_cogl_pango_font_ > map_get_priv(clutter_context_get_pango_fontmap())->renderer))" > > (this needs the debugging symbols installed). > > I don't know what, exactly, causes these stale textures, but it seems to > happen every now and then after a suspend/resume cycle. It may therefore > also be a graphics driver bug. I'm still reporting it here against > cogl-pango because this is as far as I've been able to debug the issue with > my limited knowledge of the full stack. > > I observed this issue both under the X session and under Wayland. > > Here's some version information for my system: > > Ubuntu 16.04 + gnome-staging ppa > (https://launchpad.net/~gnome3-team/+archive/ubuntu/gnome3-staging) > > ii gnome-shell 3.20.1-0ubuntu1~xenial3 > amd64 graphical shell for the GNOME desktop > ii libclutter-1.0-0:amd64 1.26.0-2~xenial0 > amd64 Open GL based interactive canvas library > ii libcogl-pango20:amd64 1.22.0-2 > amd64 Object oriented GL/GLES Abstraction/Utility > Layer > ii libdrm-intel1:amd64 2.4.67-1 > amd64 Userspace interface to intel-specific kernel > DRM services -- runtime > > Intel HD Graphics 4000 > > For reference, here's a question about this on askubuntu, together with my > answer detailing, in essence, the same as what I wrote in this bug report: > https://askubuntu.com/questions/756275/gnome-shell-text-getting-messed-up/ > 769407 > > Looking forward to any insight you may have! Can this cache be disabled somehow?
Can this cache be reloaded somehow? It would be nice to be able to overcome this without restarting machine. It happens quite often.
(In reply to umpirsky from comment #16) > Can this cache be reloaded somehow? There is no known, generic notification for us to know that the machine has been suspended or resumed.
Re-assigning to Mutter, as this involves its internal copy of Cogl. The Shell could, conceivably, drop all FBOs on suspend, since it's in the perfect position to know if that happens.
GNOME is going to shut down bugzilla.gnome.org in favor of gitlab.gnome.org. As part of that, we are mass-closing older open tickets in bugzilla.gnome.org which have not seen updates for a longer time (resources are unfortunately quite limited so not every ticket can get handled). If you can still reproduce the situation described in this ticket in a recent and supported software version, then please follow https://wiki.gnome.org/GettingInTouch/BugReportingGuidelines and create a new ticket at https://gitlab.gnome.org/GNOME/mutter/-/issues/ Thank you for your understanding and your help.