Bug 270733 – [GStreamer] MacBook Air website consistently crashes in WebCore::SourceBufferPrivateGStreamer::flush

• Status: RESOLVED
• Resolution: DUPLICATE
• Component: Media

[Caden Mitchell]

# Description

I went to Apple.com to browse for a new MacBook Air to test Asahi Linux on. Ironically, WebKit crashes when I load this webpage: https://www.apple.com/macbook-air/

# More information

I initially loaded this up in GNOME Web 45.2 (WebKitGTK 2.42.5), and it immediately crashed. Every time the webpage loaded, it would render the page perfectly fine for the first frame, become unresponsive, then crash. This was 100% reproduceable. I then tried opening it up in the Nightly version. It actually worked flawlessly on first try, no crashing. However, when I deleted the appdata folder (~/.var/app/org.gnome.Epiphany.Devel) and copied my files from the stable version of Epiphany (~/.var/app/org.gnome.Epiphany) into that folder, it had the same crashing problem. Then, I completely nuked the appdata folder, force quit all my Epiphany processes, and tested on a fresh appdata, and it exhibited the same behavior. I believe the page working the one time was a fluke related to some combination of cache or cookies left over from a past browser session... :-/

# Steps to reproduce

1. Go to https://www.apple.com/macbook-air/ on the stable and nightly versions of GNOME Web

# Debugging information:

GNOME Web 45.2 (stable):

WebKitGTK 2.42.5
GStreamer 1.22.10

GNOME Web 46.rc-11-gdaf5125aa+ (nightly)

WebKitGTK 2.43.4
GStreamer 1.22.10

[Michael Catanzaro]

Please attach a stack trace following the instructions https://handbook.gnome.org/issues/stack-traces.html

[Caden Mitchell]

Created attachment 470309 [details]
GDB stack trace for WebKitGTK webpage crash apple.com/macbook-air on 2024.03.11

This is what happens when you try to load https://apple.com/macbook-air in WebKitGTK

[Michael Catanzaro]

That is not actionable. Please reread the instructions and try again.

[Caden Mitchell]

(In reply to Michael Catanzaro from comment #3)
> That is not actionable. Please reread the instructions and try again.

It's been a while but as this issue is still present, I've since figured out how to get a backtrace so here it is:

https://gist.github.com/TriVoxel/56c399ae02368926c19710dc00c2e389

[Michael Catanzaro]

Please attach the backtrace to this bug report.

[Philippe Normand]

*** This bug has been marked as a duplicate of bug 260455 ***

[Caden Mitchell]

Created attachment 473731 [details]
Attached backtrace to the issue

Here is the back trace in attachment form as requested.