[Webkit-unassigned] [Bug 233584] [GTK] Crash in WebKit::WebsiteDataStore::fetchDataAndApply

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=233584

--- Comment #4 from Michael Catanzaro <mcatanzaro at gnome.org> ---
(In reply to LJoris from comment #1)
> this can definitely be reproduced by loading a page where animated gif are
> playing, since there is a freeze there is no coredump available

Well you posted a backtrace for the crash, so you must have gotten it from a core dump.

I wonder if you're able to reproduce the actual WebKitWebsiteDataManager crash somehow. The backtrace at https://bugs.webkit.org/show_bug.cgi?id=233578#c6 is sufficiently weird that I actually wonder whether you might have a hardware failure here (this is *very* common and, yes, very scary). It's *probably* a WebKit bug, maybe memory corruption somewhere, but it's worth running a RAM test (e.g. memtest86) just in case. Normally RAM tests fail to detect bad RAM, but if it does catch a problem, that would explain a lot.

(gdb) bt full
#0  std::__atomic_base<unsigned int>::operator++() () at /usr/include/c++/10/bits/atomic_base.h:326
#1  WTF::ThreadSafeRefCountedBase::ref() const () at WTF/Headers/wtf/ThreadSafeRefCounted.h:60
#2  WTF::Ref<WebKit::WebsiteDataStore, WTF::RawPtrTraits<WebKit::WebsiteDataStore> >::Ref(WebKit::WebsiteDataStore&) () at WTF/Headers/wtf/Ref.h:67
#3  CallbackAggregator () at ../Source/WebKit/UIProcess/WebsiteData/WebsiteDataStore.cpp:438
#4  create () at ../Source/WebKit/UIProcess/WebsiteData/WebsiteDataStore.cpp:327
#5  WebKit::WebsiteDataStore::fetchDataAndApply(WTF::OptionSet<WebKit::WebsiteDataType>, WTF::OptionSet<WebKit::WebsiteDataFetchOption>, WTF::Ref<WTF::WorkQueue, WTF::RawPtrTraits<WTF::WorkQueue> >&&, WTF::Function<void (WTF::Vector<WebKit::WebsiteDataRecord, 0ul, WTF::CrashOnOverflow, 16ul, WTF::FastMalloc>)>&&) () at ../Source/WebKit/UIProcess/WebsiteData/WebsiteDataStore.cpp:451
#6  0x00007fbd6939c26c in WebKit::WebsiteDataStore::fetchData(WTF::OptionSet<WebKit::WebsiteDataType>, WTF::OptionSet<WebKit::WebsiteDataFetchOption>, WTF::Function<void (WTF::Vector<WebKit::WebsiteDataRecord, 0ul, WTF::CrashOnOverflow, 16ul, WTF::FastMalloc>)>&&) () at ../Source/WebKit/UIProcess/WebsiteData/WebsiteDataStore.cpp:318
#7  0x00007fbd693223fc in webkit_website_data_manager_fetch() () at ../Source/WebKit/UIProcess/API/glib/WebKitWebsiteDataManager.cpp:1031

I wonder why we can't see the values of any variables on the stack, but we do see line numbers, indicating that you have successfully installed all available debuginfo. I would only expect this if running on i386 (not enough RAM there for full debuginfo). Could you post the output of 'uname -a' please?

Since the backtrace is strange, I think most likely we'll only be able to fix this one if you can really reproduce it. And since you've attached only unrelated backtraces to this issue so far, I suspect you only hit this once by luck and can't actually reproduce.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20211130/60aa1b20/attachment-0001.htm>