To be honest, the issue description is insufficient and the crash file useless. The stacktrace does not contain any debug symbols and for the "sometimes shows screen garbage before disconnecting" part, a screenshot would be needed, because the screen content is not just garbage. It has a pattern, and that is a wrong DRM format modifier. It may not be obvious for the user, but developers don't have a glass ball in front of them to see what happens on the other side of the screen, so a screenshot can be super helpful.
I found similar stacktraces in Ubuntus error tracker. Problem with Ubuntus error tracker is that a lot stacktraces cannot be retrieved and therefore leave only a single line of calls, where only the library name is shown. And this is also here the case. However, the issue could be fixed upstream and is fixed in the current master branch of g-r-d, due to kind help of a person in https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/issues/96. Personally, I don't have an arm64 device, which is why this issue could never been detected. And without that help in form of testing patches and providing the stacktrace with debug symbols, the issue would still exist. The crash itself is kinda interesting, because it appears, that some instructions are completed in a different order on arm64. It was fixed in https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/commit/bdc8d63bddf2ebb0da97e9277267b7398e49710e and https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/commit/b7b0597ba59072b718a78760c2732e64e080b4bf. Then, to the part, what is here referred to as 'garbled content': A screenshot for this is provided here: https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/issues/96#note_1435877 Without the screenshot, I wouldn't been able to have that modifier guess. So, please, for future reports provide screenshots for such things. The issue here is that g-r-d always expected explicit DRM format modifiers. However, it is an implicit modifier here. There is no such flag to indicate that and as a result the `DRM_FORMAT_MOD_INVALID` modifier is used to indicate an implicit modifier. It is a weird thing though: In a screencast, submitting this specific modifier means, that an implicit modifier is used, while graphics APIs use this to indicate an error. However, using `DRM_FORMAT_MOD_INVALID` as modifier was never a problem with other drivers, like Intel and so it was unnoticed. The issue was fixed in https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/commit/60244e96333fbeabfee9c1a0e5ce56590dbb313d and https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/commit/d9923317f0b60621f0aeb219f4d10612a2178e63. I'll ask Jonas to push another release tag (might be 42.1.1). 42.1 was pushed saturday, but does not contain these fixes. Personally, I see this issue as severe, but I am not one, who can push tags in g-r-d. So, I can only ask Jonas to push one. In any case, Ubuntu can also build g-r-d from the master branch until commit https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/commit/d9923317f0b60621f0aeb219f4d10612a2178e63 to push the fix to Ubuntu here. ** Bug watch added: gitlab.gnome.org/GNOME/gnome-remote-desktop/-/issues #96 https://gitlab.gnome.org/GNOME/gnome-remote-desktop/-/issues/96 -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1970139 Title: VNC Connection doesn't work on arm64 (Raspberry Pi 4 8Gb) in Ubuntu 22.04 LTS To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/gnome-remote-desktop/+bug/1970139/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
