Description of problem: Set org.gnome.desktop.interface font-name to an empty string with dconf-editor, then try to load a web page. Upstream bug incoming.... Version-Release number of selected component: webkitgtk4-2.8.3-3.fc22 Additional info: reporter: libreport-2.5.1 backtrace_rating: 4 cmdline: /usr/libexec/webkit2gtk-4.0/WebKitWebProcess 20 crash_function: rawHash executable: /usr/libexec/webkit2gtk-4.0/WebKitWebProcess global_pid: 27183 kernel: 4.0.4-303.fc22.x86_64 runlevel: N 5 type: CCpp uid: 1000 var_log_messages: [System Logs]:\n-- Logs begin at Tue 2015-05-19 11:54:49 CDT, end at Tue 2015-06-23 15:35:58 CDT. -- Truncated backtrace: Thread no. 1 (10 frames) #0 rawHash at /usr/src/debug/webkitgtk-2.8.3/Source/WTF/wtf/text/StringImpl.h:500 #1 hasHash at /usr/src/debug/webkitgtk-2.8.3/Source/WTF/wtf/text/StringImpl.h:506 #2 hash at /usr/src/debug/webkitgtk-2.8.3/Source/WTF/wtf/text/StringImpl.h:517 #5 hash<std::pair<WTF::String, bool> > at /usr/src/debug/webkitgtk-2.8.3/Source/WTF/wtf/HashMap.h:171 #6 add<WTF::HashMapTranslator<WTF::HashMap<std::pair<WTF::String, bool>, WTF::RefPtr<WebCore::CSSPrimitiveValue> >::KeyValuePairTraits, WTF::PairHash<WTF::String, bool> >, std::pair<WTF::String, bool>, std::nullptr_t> at /usr/src/debug/webkitgtk-2.8.3/Source/WTF/wtf/HashTable.h:807 #7 inlineAdd<std::pair<WTF::String, bool>, std::nullptr_t> at /usr/src/debug/webkitgtk-2.8.3/Source/WTF/wtf/HashMap.h:296 #8 WTF::HashMap<std::pair<WTF::String, bool>, WTF::RefPtr<WebCore::CSSPrimitiveValue>, WTF::PairHash<WTF::String, bool>, WTF::HashTraits<std::pair<WTF::String, bool> >, WTF::HashTraits<WTF::RefPtr<WebCore::CSSPrimitiveValue> > >::add<decltype(nullptr)>(std::pair<WTF::String, bool>&&, decltype(nullptr)&&) at /usr/src/debug/webkitgtk-2.8.3/Source/WTF/wtf/HashMap.h:331 #9 WebCore::CSSValuePool::createFontFamilyValue at /usr/src/debug/webkitgtk-2.8.3/Source/WebCore/css/CSSValuePool.cpp:130 #10 WebCore::CSSParser::parseSystemFont at /usr/src/debug/webkitgtk-2.8.3/Source/WebCore/css/CSSParser.cpp:6356 #11 WebCore::CSSParser::parseValue at /usr/src/debug/webkitgtk-2.8.3/Source/WebCore/css/CSSParser.cpp:2932 Potential duplicate: bug 1186561
Created attachment 1042493 [details] File: backtrace
Created attachment 1042494 [details] File: cgroup
Created attachment 1042495 [details] File: core_backtrace
Created attachment 1042496 [details] File: dso_list
Created attachment 1042497 [details] File: environ
Created attachment 1042498 [details] File: limits
Created attachment 1042499 [details] File: maps
Created attachment 1042500 [details] File: mountinfo
Created attachment 1042501 [details] File: namespaces
Created attachment 1042502 [details] File: open_fds
Created attachment 1042503 [details] File: proc_pid_status
Another user experienced a similar problem: 1. Open "duckduckgo.com". 2. Right click anywhere in the page, then choose "Inspect element". 3. With the "Web Inspector" still opened, click the reload button. reporter: libreport-2.6.0 backtrace_rating: 4 cmdline: /usr/libexec/webkit2gtk-4.0/WebKitWebProcess 23 crash_function: rawHash executable: /usr/libexec/webkit2gtk-4.0/WebKitWebProcess global_pid: 11619 kernel: 4.0.6-300.fc22.x86_64 package: webkitgtk4-2.8.3-2.fc22 reason: WebKitWebProcess killed by SIGSEGV runlevel: N 5 type: CCpp uid: 1000
Created attachment 1046106 [details] Screencast of the steps previously mentioned. Always happen. Not happen with "google.com" page, btw. ABRT lead me to here. I don't know what to do :P Not sure if really a duplicate. In the worst case scenario, this new information should serve as a test case to the upstream bug, I think.
(In reply to Diogo Campos from comment #12) > Another user experienced a similar problem: > > 1. Open "duckduckgo.com". > 2. Right click anywhere in the page, then choose "Inspect element". > 3. With the "Web Inspector" still opened, click the reload button. Strange, I can't reproduce this. Surely it is a different bug and ABRT got this wrong.
I checked a bit more: seems weirdly specific. - It needs to be in the search results page. (tested exactly with "https://duckduckgo.com/?q=how+to+make+this+thing+work&t=epiphany"). *AND* - The "region switch" needs to be turned on. (tested to happen with "Brazil", "Argentina", "Spain", "United Kingdom"; but *doesn't* happen with "United States"). So: make sure this two conditions are met in "Step 1", then do "Step 2" and "Step 3".
OK, now I can reproduce it. Thanks; that is a weirdly-specific indeed. And the UI process does not recover the web view properly after the crash; it just remains white forever.