Firefox regularly freezes when using autoscroll
Categories
(Core :: Widget: Gtk, defect, P2)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox146 | --- | fixed |
People
(Reporter: gwarser, Assigned: stransky)
References
Details
Attachments
(1 obsolete file)
User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:146.0) Gecko/20100101 Firefox/146.0
Steps to reproduce:
During browsing when using autoscroll activelly firefox randomly freezes and I need to close it (right click on task bar). Last time happened on https://github.com/gorhill/uBlock/releases/tag/1.67.1b2 Crash was generated https://crash-stats.mozilla.org/report/index/7c7b3fe9-00ce-4563-be1f-3783d0251102#tab-details
I believe this started with version 20251028215909
I'm on Manjaro KDE
Comment 1•6 months ago
|
||
The Bugbug bot thinks this bug should belong to the 'Core::Widget: Gtk' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.
| Assignee | ||
Comment 2•6 months ago
|
||
The crash looks weird, like memory error or so. Please check and attach more crashes from about:crashes if you have any.
Also please try plain Mozilla binaries:
https://fedoraproject.org/wiki/How_to_debug_Firefox_problems#Testing_Mozilla_binaries
Thanks.
Comment 3•6 months ago
|
||
I'm also experiencing this, but not sure how to reproduce consistently, so can't investigate with mozregression.
How can I generate a crash report if the browser freezes?
I'm using Mozilla binaries.
https://crash-stats.mozilla.org/report/index/43f9db82-876a-40f2-88b1-af6e80251102
https://crash-stats.mozilla.org/report/index/57546d98-7b55-483d-a5ef-f4e9b0251102
https://crash-stats.mozilla.org/report/index/4639f6f6-602b-450f-9e57-ae1460251102
https://crash-stats.mozilla.org/report/index/09bda7f9-730c-4169-b2d2-f20000251102
https://crash-stats.mozilla.org/report/index/e5f40fce-2f2f-4ffb-baa0-7712d0251102
https://crash-stats.mozilla.org/report/index/b1e38e62-f777-40ea-9063-945650251102
https://crash-stats.mozilla.org/report/index/2838e7d5-2ce9-4cef-8f1f-00a740251102
https://crash-stats.mozilla.org/report/index/81a926fd-c24b-4daf-ada7-7c0540251102
https://crash-stats.mozilla.org/report/index/7c7b3fe9-00ce-4563-be1f-3783d0251102
https://crash-stats.mozilla.org/report/index/e29da7a8-5f5b-4914-8475-f8dca0251101
https://crash-stats.mozilla.org/report/index/4c17e19d-9eb8-4859-b823-078010251101
I will try using kill 11 next time.
kill -s SIGSEGV did not seem to change anything:
https://crash-stats.mozilla.org/report/index/0c5f71b5-7da9-430a-88b8-eedd50251103
https://crash-stats.mozilla.org/report/index/7e11d925-5d3a-4933-8ddc-ff0880251103
Comment 6•6 months ago
|
||
I can confirm this is happening when autoscrolling and only when autoscrolling. It has crashed probably 7 times today from this. The crashes freeze firefox and must be sent a kill signal to stop it. No crash reports that I can see are made or are able to be submitted.
Archlike linux up to date as of yesterday and on 146.0a1 (2025-11-02) (64-bit).
Plasma KDE 6.5.1, Wayland.
The only thing I saw in the journal was these lines:
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: pci id for fd 30: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 29: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 31: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: egl: failed to create dri2 screen
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: pci id for fd 30: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 29: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 31: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: egl: failed to create dri2 screen
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: pci id for fd 30: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: pci id for fd 30: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 29: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 31: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: egl: failed to create dri2 screen
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: DRI2: failed to create screen
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: pci id for fd 30: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 29: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: pci id for fd 31: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: egl: failed to create dri2 screen
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: DRI2: failed to create screen
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: libEGL warning: pci id for fd 30: 10de:21c4, driver (null)
Nov 03 10:07:16 cachyos-x8664 firefox[283480]: [GFX1-]: DMABufSurface::UseDmaBufExportExtension(): MESA_image_dma_buf import/export extensions!
and maybe this but it's not something that happened for every crash:
Nov 03 12:47:50 cachyos-x8664 firefox[394898]: [Child 394898, MediaDecoderStateMachine #1] WARNING: Decoder=7f6b97fa9400 Decode error: NS_ERROR_ABORT (0x80004004) - Result<already_AddRefed<MediaRawData>, MediaResult> mozilla::SampleIterator::GetNext(): Sample data read failed: file /builds/worker/checkouts/>
Updated•6 months ago
|
I have the same issue (Firefox nightly on Arch Linux, Gnome 49 / Wayland), I don't know exactly when it started by I think it was a few weeks ago.
It does not happen very often (once every multiple hours).
Each time I killed Firefox with the "Force quit" button that Gnome shows when programs are unresponsive.
From what I saw, once that happens Firefox entirely stops responding (even when multiples windows are open), but I think sounds continues normally (not entirely sure about that).
I also think that it only happens when (not 100% sure):
- Autoscroll is active
- I click to stop it.
Unfortunately I am unable to reproduce it reliably, and Firefox never generated a crash report when that happened, but I guess this is because the Gnome sends SIGKILL not giving opportunity to Firefox to create a log.
How can I help to debug that (e.g. running Firefox through a debugger)?
Well that's funny but I had the issue 30 seconds after posting my message :D
So I was able to try some things: the profile manager opens but does not work when I click on "Start Nightly", even if I change the selected profile (no window opens, although new processes are created).
I cannot resize windows, and shortcuts like "New private window" do nothing.
I sent the signal SIGSEGV on all processes named "firefox-bin" by using the command "killall -SIGSEGV -e firefox-bin".
It generated two crash reports (maybe because of the profile manager test), I don't know if they are exploitable:
https://crash-stats.mozilla.org/report/index/bp-4a03a5e1-fc09-4a25-ae91-b2d140251103
https://crash-stats.mozilla.org/report/index/bp-2e5b775a-88bb-43e2-9325-335200251103
| Assignee | ||
Comment 9•6 months ago
|
||
From the log it looks like we're blocked on window close while compositor thread tries to finish rendering:
6 libxul.so mozilla::layers::PCompositorBridgeChild::SendWillClose() ipc/ipdl/PCompositorBridgeChild.cpp:596 inlined
6 libxul.so mozilla::layers::CompositorBridgeChild::SendWillClose() gfx/layers/ipc/CompositorBridgeChild.cpp:313 inlined
6 libxul.so mozilla::layers::CompositorBridgeChild::Destroy() gfx/layers/ipc/CompositorBridgeChild.cpp:180 cfi
7 libxul.so mozilla::layers::InProcessCompositorSession::Shutdown() gfx/ipc/InProcessCompositorSession.cpp:99 cfi
8 libxul.so nsIWidget::DestroyCompositor() widget/nsIWidget.cpp:377 cfi
9 libxul.so nsWindow::OnUnmap() widget/gtk/nsWindow.cpp:10097 cfi
10 libxul.so moz_container_unmap(_GtkWidget*)
compositor:
0 libc.so.6 futex_wait /usr/src/debug/glibc/glibc/sysdeps/nptl/futex-internal.h:146 inlined
0 libc.so.6 __GI___lll_lock_wait /usr/src/debug/glibc/glibc/nptl/lowlevellock.c:49 context
1 libc.so.6 lll_mutex_lock_optimized /usr/src/debug/glibc/glibc/nptl/pthread_mutex_lock.c:48 inlined
1 libc.so.6 ___pthread_mutex_lock /usr/src/debug/glibc/glibc/nptl/pthread_mutex_lock.c:128 cfi
2 libxul.so mozilla::RecursiveMutex::LockInternal() xpcom/threads/RecursiveMutex.cpp:71 inlined
2 libxul.so mozilla::RecursiveMutex::Lock() xpcom/threads/RecursiveMutex.h:31 inlined
2 libxul.so mozilla::RecursiveMutexAutoLock::RecursiveMutexAutoLock(mozilla::RecursiveMutex&) xpcom/threads/RecursiveMutex.h:80 inlined
2 libxul.so mozilla::layers::AsyncPanZoomController::RequestContentRepaint(mozilla::layers::RepaintRequest::ScrollOffsetUpdateType)
a solution may be to make the nsWindow::OnUnmap() / nsIWidget::DestroyCompositor() non-blocking on main thread - Bug 1739232.
Comment 10•6 months ago
|
||
Not sure how helpful it is since you have all those other reports but here are two crashes in a brand new profile:
https://crash-stats.mozilla.org/report/index/95392b92-f35d-4963-8c84-6c57b0251105
https://crash-stats.mozilla.org/report/index/5863594d-8a22-45c9-a2f1-1f6ee0251105
It's fairly easy to cause the crash if you press the middle mouse button a lot and scroll/cancel the scroll.
| Assignee | ||
Comment 11•5 months ago
|
||
Linux used a workaround for Bug 1654938 but it causes browser hang/freezes so let's remove it and follow other platforms.
Updated•5 months ago
|
| Assignee | ||
Updated•5 months ago
|
| Assignee | ||
Updated•5 months ago
|
Comment 12•5 months ago
|
||
This crash was fixed in bug 1998057.
Comment 13•5 months ago
|
||
Comment on attachment 9528949 [details]
Bug 1997782 [Linux] Don't delete layer manager for hidden popups r?emilio
Revision D273965 was moved to bug 2002206. Setting attachment 9528949 [details] to obsolete.
Updated•5 months ago
|
Description
•