Bug 164919
| Summary: | [GTK] Crash in WebCore::TextureMapperShaderProgram in secondary thread while _mesa_destroy_shader_compiler is running in exit handler | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Michael Catanzaro <mcatanzaro> |
| Component: | WebKitGTK | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED DUPLICATE | ||
| Severity: | Normal | CC: | bugs-noreply, cgarcia, magomez, mcatanzaro, zan |
| Priority: | P2 | ||
| Version: | WebKit Nightly Build | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| See Also: | https://bugzilla.redhat.com/show_bug.cgi?id=1383032 | ||
Michael Catanzaro
Crash in WebCore::TextureMapperShaderProgram in secondary thread while _mesa_destroy_shader_compiler is running in exit handler. I have no truncated backtrace for this, just check the full backtrace in the downstream bug, threads 1 and 2. I guess either WebCore::TextureMapperShaderProgram is doing something that is not threadsafe, or maybe it needs to install its own exit handler.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Carlos Garcia Campos
I remember this, I think it was that we were allocating an AtomicString in a thread and releasing it in a different thread. I think Zan fixed this.
#0 0x00007fcaae1e2cba in WTF::HashTable<WTF::AtomicString, WTF::KeyValuePair<WTF::AtomicString, unsigned int>, WTF::KeyValuePairKeyExtractor<WTF::KeyValuePair<WTF::AtomicString, unsigned int> >, WTF::AtomicStringHash, WTF::HashMap<WTF::AtomicString, unsigned int>::KeyValuePairTraits, WTF::HashTraits<WTF::AtomicString> >::add<WTF::HashMapEnsureTranslator<WTF::HashMap<WTF::AtomicString, unsigned int>::KeyValuePairTraits, WTF::AtomicStringHash>, const WTF::AtomicString&, const WebCore::TextureMapperShaderProgram::getLocation(const WTF::AtomicString&, WebCore::TextureMapperShaderProgram::VariableType)::<lambda()>&> (extra=..., key=..., this=0x7fca97d90428) at /usr/src/debug/webkitgtk-2.14.0/Source/WTF/wtf/HashTable.h:877
877 unsigned sizeMask = m_tableSizeMask;
Michael Catanzaro
Looks like Zan told me how to fix this... so kinda right!
*** This bug has been marked as a duplicate of bug 161800 ***