Telegram Desktop crashing (Linux, Flatpak, GNOME, Wayland, 2 screens)

[Himura2la]

Steps to reproduce

1. Use the client normally in two-screens environment: chat, view media, etc. Media viewer loads on the same screen with the main window.
2. Switch between workspaces normally. After the first switch, media viewer starts opening on a second screen.
3. Suddenly, Telegram crashes right after launching media viewer on a random image. Some images seems safe, some may crash on 5-6 iteration of workspace change + image opening. But I'm not sure that something depends on the image, it's stochastic.

example of the image that crashed me several times: https://t.me/rbtshki/26768

Expected behaviour

No crashes

Actual behaviour

Telegram may suddenly crash on opening a media viewer with the free(): invalid pointer line in flatpak console.

Operating system

RHEL 10

Version of Telegram Desktop

6.3.6

Installation source

Static binary from official website

Crash ID

No response

Logs

$ flatpak run org.telegram.desktop 
qt.waylandcompositor.hardwareintegration: Failed to initialize EGL display. There is no EGL_WL_bind_wayland_display extension.
qt.waylandcompositor.hardwareintegration: Failed to initialize EGL display. There is no EGL_WL_bind_wayland_display extension.
QPainter::begin: Paint device returned engine == 0, type: 2
QWidget::render: Cannot render with an inactive painter
QDBusTrayIcon encountered a D-Bus error: QDBusError("org.freedesktop.DBus.Error.ServiceUnknown", "The name is not activatable")
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
[libvpx-vp9 @ 0x7f4287340d80] v1.15.2
[libvpx-vp9 @ 0x7f427c0fffc0] v1.15.2
[libvpx-vp9 @ 0x7f428891dc00] v1.15.2
[libvpx-vp9 @ 0x7f4278a20380] v1.15.2
[libvpx-vp9 @ 0x7f4287340d80] v1.15.2
[libvpx-vp9 @ 0x7f428891dc00] v1.15.2
[libvpx-vp9 @ 0x7f4278a20380] v1.15.2
[libvpx-vp9 @ 0x7f427c0fffc0] v1.15.2
[libvpx-vp9 @ 0x7f428747e300] v1.15.2
[libvpx-vp9 @ 0x7f428747e300] v1.15.2
[libvpx-vp9 @ 0x7f429450a640] v1.15.2
[libvpx-vp9 @ 0x7f429450a640] v1.15.2
[libvpx-vp9 @ 0x7f428800b740] v1.15.2
[libvpx-vp9 @ 0x7f42805f4dc0] v1.15.2
[libvpx-vp9 @ 0x7f428800b740] v1.15.2
[libvpx-vp9 @ 0x7f42805f4dc0] v1.15.2
[libvpx-vp9 @ 0x7f4280fb6c80] v1.15.2
[libvpx-vp9 @ 0x7f428e289a40] v1.15.2
[libvpx-vp9 @ 0x7f428e243000] v1.15.2
[libvpx-vp9 @ 0x7f4280fb6c80] v1.15.2
[libvpx-vp9 @ 0x7f42940309c0] v1.15.2
[libvpx-vp9 @ 0x7f428e289a40] v1.15.2
[libvpx-vp9 @ 0x7f428e243000] v1.15.2
[libvpx-vp9 @ 0x7f42940309c0] v1.15.2
[libvpx-vp9 @ 0x7f4288243f80] v1.15.2
[libvpx-vp9 @ 0x7f4288243f80] v1.15.2
free(): invalid pointer

$ flatpak run org.telegram.desktop 
qt.waylandcompositor.hardwareintegration: Failed to initialize EGL display. There is no EGL_WL_bind_wayland_display extension.
qt.waylandcompositor.hardwareintegration: Failed to initialize EGL display. There is no EGL_WL_bind_wayland_display extension.
QPainter::begin: Paint device returned engine == 0, type: 2
QWidget::render: Cannot render with an inactive painter
QDBusTrayIcon encountered a D-Bus error: QDBusError("org.freedesktop.DBus.Error.ServiceUnknown", "The name is not activatable")
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
[libvpx-vp9 @ 0x7f7664087440] v1.15.2
[libvpx-vp9 @ 0x7f767048c3c0] v1.15.2
[libvpx-vp9 @ 0x7f7664087440] v1.15.2
[libvpx-vp9 @ 0x7f767048c3c0] v1.15.2
[libvpx-vp9 @ 0x7f7674da0ec0] v1.15.2
[libvpx-vp9 @ 0x7f766875d200] v1.15.2
[libvpx-vp9 @ 0x7f7661022200] v1.15.2
[libvpx-vp9 @ 0x7f765c349640] v1.15.2
[libvpx-vp9 @ 0x7f7674da0ec0] v1.15.2
[libvpx-vp9 @ 0x7f766875d200] v1.15.2
[libvpx-vp9 @ 0x7f765c349640] v1.15.2
[libvpx-vp9 @ 0x7f7661022200] v1.15.2
[libvpx-vp9 @ 0x563548952e40] v1.15.2
[libvpx-vp9 @ 0x563548952e40] v1.15.2
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.
free(): invalid pointer

[Aokromes]

plz enable beta, restart telegram and send crash id on next start and post id here.

[larsblumberg]

Related: #30046

[ilya-fedin]

plz enable beta, restart telegram and send crash id on next start and post id here.

And log.txt.

[Himura2la]

There's nothing in log.txt about the crash, I've checked. What about the crash id, I'm afraid I can't bypass selinux and allow core dumps, but I'll try to upload the crash dump from the other device later. It also reproduces on NixOS

[Himura2la]

BTW, I don't use NVIDIA, my hardware is Acer Swift Go 16 SFG16-72T

[ilya-fedin]

There's nothing in log.txt about the crash, I've checked.

Provide it anyway. It's a requirement.

[Himura2la]

done

[Himura2la]

I did not find how to enable beta on flatpack ( Also, I don't see AVC denials anymore after crash, so I guess crashes are not processed...

[ilya-fedin]

You can't, you have to reproduce on static binary

[kukhalex]

Hey guys,

Did anybody resolved this issue?

I have the same problem.

Logs saying:

QIBusPlatformInputContext: invalid portal bus.

qt.waylandcompositor.hardwareintegration: Failed to initialize EGL display. There is no EGL_WL_bind_wayland_display extension.

qt.waylandcompositor.hardwareintegration: Failed to initialize EGL display. There is no EGL_WL_bind_wayland_display extension.

QPainter::begin: Paint device returned engine == 0, type: 2

QWidget::render: Cannot render with an inactive painter

[libvpx-vp9 @ 0x7ef4a0209fc0] v1.15.2

[libvpx-vp9 @ 0x7ef4a846d640] v1.15.2

[libvpx-vp9 @ 0x7ef4b0105a80] v1.15.2

[libvpx-vp9 @ 0x7ef4a0209fc0] v1.15.2

[libvpx-vp9 @ 0x7ef49c1f06c0] v1.15.2

[libvpx-vp9 @ 0x7ef49c1f06c0] v1.15.2

[libvpx-vp9 @ 0x7ef4b80b3040] v1.15.2

[libvpx-vp9 @ 0x7ef4b80b3040] v1.15.2

qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.

[libvpx-vp9 @ 0x7ef4b838eac0] v1.15.2

[libvpx-vp9 @ 0x7ef4b838eac0] v1.15.2

qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.

[libvpx-vp9 @ 0x7ef4b4c37500] v1.15.2

[libvpx-vp9 @ 0x7ef4b4c37500] v1.15.2

[libvpx-vp9 @ 0x7ef4a008bd40] v1.15.2

[libvpx-vp9 @ 0x7ef4a008bd40] v1.15.2

qt.gui.imageio.jpeg: ./lib/jpegli/decode_scan.cc:535: Incomplete scan detected.

The Wayland connection experienced a fatal error: Protocol error

[ilya-fedin]

@kukhalex you have a different issue according to the last line

[Hursofid]

Crashes for me too. It was working perfectly but then after clicking update in the app it stopped working. Plain binary from https://td.telegram.org/tlinux/tsetup.6.3.7.tar.xz, Debian 12, Gnome on X11 Tried two previous versions, but it still crashes. "Last time Telegram Desktop was not closed properly" and that's it.

[54192.203972] Telegram[23191]: segfault at 8 ip 0000560b1ba8f6bc sp 00007ffeef25ba10 error 4 in Telegram[560b15810000+7259000] likely on CPU 2 (core 2, socket 0)
[54192.203985] Code: 4c 89 c7 ff e0 90 0f 1f 84 00 00 00 00 00 f3 0f 1e fa 55 48 89 e5 41 56 41 55 41 89 cd 41 54 49 89 f4 53 48 89 fb 48 83 ec 10 <4c> 8b 32 64 48 8b 04 25 28 00 00 00 48 89 45 d8 31 c0 4d 85 f6 74

[frgigr]

Same issue on Debian 12, using the latest standalone Telegram Desktop executable. This is the error message from terminal, hope it helps

(Telegram:4942): GLib-CRITICAL **: 11:53:20.031: g_unix_open_pipe: assertion '(flags & (FD_CLOEXEC)) == flags' failed
Segmentation Fault

[ilya-fedin]

@Hursofid @frgigr no, it doesn't help, please read the issue, it's about a completely different bug

[Hursofid]

@Hursofid @frgigr no, it doesn't help, please read the issue, it's about a completely different bug

I was trying to not create more issues for devs. In regards, will create a separate one.

[ilya-fedin]

@Hursofid don't create, there's already an issue for your case, it's just not this one

[Hursofid]

@Hursofid don't create, there's already an issue for your case, it's just not this one

Yeah, I've realised that. It appeared after I've drafted and sent my comment, so I'm not creating new one