159919 2016-07-19 06:09:18 -0700 Not possible to GDB WebKitGtk+ 2.13.3: demangler-warning: unable to demangle 2016-09-17 08:11:00 -0700 1 1 1 Unclassified WebKit WebKitGTK WebKit Nightly Build Unspecified Unspecified RESOLVED INVALID https://bugzilla.redhat.com/show_bug.cgi?id=1377020 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68159 P2 Normal --- 1 agomez webkit-unassigned aperez bugs-noreply clopez fujii jfernandez mcatanzaro pnormand oldest_to_newest 1212337 0 agomez 2016-07-19 06:09:18 -0700 I'm using WebKitGtk+ with my own JHBuild setting: https://github.com/tanty/jhbuild-epiphany/tree/master Epiphany 3.20.3 and WebKit 2.13.3 I'm running Epiphany with the dconf key: "process-model" = "shared-secondary-process" The compilation was done with CMake args: '-DPORT=GTK -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_FLAGS_RELEASE="-O0 -g1 -DNDEBUG -DG_DISABLE_CAST_CHECKS" -DCMAKE_CXX_FLAGS_RELEASE="-O0 -g1 -DNDEBUG -DG_DISABLE_CAST_CHECKS"' When trying to attach with gdb to any of the running processes (epiphany, WebKitNetworkProcess or WebKitWebProcess), I get: /build/gdb-7.10/gdb/cp-support.c:1595: demangler-warning: unable to demangle '_ZSt7forwardIRZN6WebKit29NetworkConnectionToWebProcess26writeBlobsToTemporaryFilesERKN3WTF6VectorINS2_6StringELm0ENS2_15CrashOnOverflowELm16EEEmEUlRT_E_EOS9_RNSt16remove_referenceIS9_E4typeE' (demangler failed with signal 11) And the only option is to exit the debugging session. 1212338 1 agomez 2016-07-19 06:11:10 -0700 Compilation and gdb done in current Debian Testing: $ gdb --version GNU gdb (Debian 7.10-1.1) 7.10 $ gcc --version gcc (Debian 5.4.0-4) 5.4.0 20160609 1212346 2 agomez 2016-07-19 06:20:05 -0700 As a workaround in the meanwhile, it is possible to add: set demangle-style none Into ~/.gdbinit 1212348 3 agomez 2016-07-19 06:25:13 -0700 Reproducible also with: $ gdb --version GNU gdb (Debian 7.11.1-2) 7.11.1 1212865 4 mcatanzaro 2016-07-20 13:43:38 -0700 This is really annoying, but it's clearly a toolchain bug, not a WebKit issue. It's been reported with several different versions of gcc, so we can safely say it's a gdb issue. 1213068 5 pnormand 2016-07-21 00:11:15 -0700 Possibly related to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68383 1215101 6 clopez 2016-07-27 17:16:07 -0700 I think the issue is this one: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61321 I'm able to reproduce the issue with this test case that comes from <https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68159#c4>: $ echo 'void _ZSt7forwardIRKZN5Write14DataMapGrammarISt20back_insert_iteratorISsEEC4EvEUlRT_E_EOS5_RNSt16remove_referenceIS5_E4typeE() {} int main() {}' | gcc -xc - -o /tmp/a.out && gdb /tmp/a.out [...] Reading symbols from /tmp/a.out.../build/gdb-GtCqbH/gdb-7.11.1/gdb/cp-support.c:1595: demangler-warning: unable to demangle '_ZSt7forwardIRKZN5Write14DataMapGrammarISt20back_insert_iteratorISsEEC4EvEUlRT_E_EOS5_RNSt16remove_referenceIS5_E4typeE' (demangler failed with signal 11) Unable to dump core, use `ulimit -c unlimited' before executing GDB next time. /build/gdb-GtCqbH/gdb-7.11.1/gdb/cp-support.c:1608: demangler-warning: unable to demangle '_ZSt7forwardIRKZN5Write14DataMapGrammarISt20back_insert_iteratorISsEEC4EvEUlRT_E_EOS5_RNSt16remove_referenceIS5_E4typeE' (demangler failed with signal 11) 1215659 7 clopez 2016-07-29 12:34:13 -0700 The bug is definitive this one: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68159 and is still present in last GDB git master. I tried using lldb (LLVM debugger) and its also crashing (segfaults)... So... at least on my environment I'm complete unable to use a debugger on current trunk of WebKit with Debug builds. Is anybody here able to work with gdb/llvdb and debug builds? 1216222 8 clopez 2016-08-01 12:40:17 -0700 (In reply to comment #7) > The bug is definitive this one: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68159 and is still present in > last GDB git master. > > I tried using lldb (LLVM debugger) and its also crashing (segfaults)... > > > So... at least on my environment I'm complete unable to use a debugger on > current trunk of WebKit with Debug builds. > > > Is anybody here able to work with gdb/llvdb and debug builds? BTW, adrian perez pointed me to this very helpful trick: * Put in your ~/.gdbinit file: set demangle-style none * Then pipe the output of GDB through c++filt (a stand-alone demangler). It works for me. I see that the WebKit tooling for generating GDB crash logs is already piping the output through c++filt. So I'm going to just set this line on the GTK+ bots ~/.gdbinit files.