-
ggaren@apple.com authored
2008-05-27 Geoffrey Garen <ggaren@apple.com> Reviewed by Tim Hatcher. Fixed https://bugs.webkit.org/show_bug.cgi?id=19183 REGRESSION (r33979): Crash in DebuggerCallFrame::functionName when clicking button in returnEvent-crash.html Added two new debugger hooks, willExecuteProgram and didExecuteProgram, along with code to generate them, code to invoke them when unwinding due to an exception, and code to dump them. SunSpider reports no change. * VM/CodeBlock.cpp: (KJS::debugHookName): I had to mark this function NEVER_INLINE to avoid a .4% performance regression. The mind boggles. WebCore: 2008-05-27 Geoffrey Garen <ggaren@apple.com> Reviewed by Tim Hatcher. Fixed https://bugs.webkit.org/show_bug.cgi?id=19183 REGRESSION (r33979): Crash in DebuggerCallFrame::functionName when clicking button in returnEvent-crash.html Added implementations for willExecuteProgram and didExecuteProgram. They take care to update our call frame when entering and exiting programs, preventing us from keeping around a stale global frame after executing a program. eval programs now show up as "anonymous function" in a new scope. This is slightly better than what they used to do -- overwriting the current scope -- but obviously we can do better. WebKit/mac: 2008-05-27 Geoffrey Garen <ggaren@apple.com> Reviewed by Tim Hatcher. Fixed https://bugs.webkit.org/show_bug.cgi?id=19183 REGRESSION (r33979): Crash in DebuggerCallFrame::functionName when clicking button in returnEvent-crash.html Added implementations for willExecuteProgram and didExecuteProgram, which take care of making sure we're not hanging on to stale data. git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34182 268f45cc-cd09-0410-ab3c-d52691b4dbfc
d0740c87