Closed
Bug 657587
Opened 13 years ago
Closed 13 years ago
TI: Crash [@ DoMatch]
Categories
(Core :: JavaScript Engine, defect)
Tracking
()
RESOLVED
WORKSFORME
People
(Reporter: decoder, Unassigned)
References
Details
(Keywords: crash, testcase)
Crash Data
Attachments
(1 file)
1.63 KB,
application/x-compressed-tar
|
Details |
The attached testcase crashes on optimized TI revision 0cd135eb71a3 (run either with -m -n or with m), tested on 64 bit. Interestingly, the test crashes instantly with -m only and takes a few seconds with -m -n. Backtrace with -m -n: ==30374== Invalid read of size 8 ==30374== at 0x526711: DoMatch(JSContext*, js::RegExpStatics*, js::Value*, JSString*, RegExpPair const&, bool (*)(JSContext*, js::RegExpStatics*, unsigned long, void*), void*, MatchControlFlags) (jsstr.h:250) ==30374== by 0x52C4E2: js::str_replace(JSContext*, unsigned int, js::Value*) (jsstr.cpp:2406) ==30374== by 0x6307AC: CallCompiler::generateNativeStub() (jscntxtinlines.h:293) ==30374== by 0x62D2F2: js::mjit::ic::NativeCall(js::VMFrame&, js::mjit::ic::CallICInfo*) (MonoIC.cpp:1108) ==30374== by 0xB6A1C43: ??? ==30374== by 0x5C658D: js::mjit::EnterMethodJIT(JSContext*, js::StackFrame*, void*, js::Value*) (MethodJIT.cpp:882) ==30374== by 0x5C679C: js::mjit::JaegerShotAtSafePoint(JSContext*, void*) (MethodJIT.cpp:912) ==30374== by 0x67EB3D: js::Interpret(JSContext*, js::StackFrame*, unsigned int, js::InterpMode) (jsinterp.cpp:6463) ==30374== by 0x646BDE: js_InternalInterpret (InvokeHelpers.cpp:1627) ==30374== by 0x5C5F6D: ??? (in /home/decoder/LangFuzz/jaegermonkey/js/src/shell/js) ==30374== by 0x5C658D: js::mjit::EnterMethodJIT(JSContext*, js::StackFrame*, void*, js::Value*) (MethodJIT.cpp:882) ==30374== by 0x5C6E65: js::mjit::JaegerShot(JSContext*) (MethodJIT.cpp:912) ==30374== Address 0x5 is not stack'd, malloc'd or (recently) free'd ==30374== ==30374== ==30374== Process terminating with default action of signal 11 (SIGSEGV) Backtrace with -m: ==31534== Invalid read of size 1 ==31534== at 0x5CB6BB: js::mjit::Compiler::restoreVarType() (Compiler.cpp:6912) ==31534== by 0x5E49BF: js::mjit::Compiler::generateMethod() (Compiler.cpp:2217) ==31534== by 0x5E8C8A: js::mjit::Compiler::performCompilation(js::mjit::JITScript**) (Compiler.cpp:531) ==31534== by 0x5E8F79: js::mjit::TryCompile(JSContext*, js::StackFrame*) (Compiler.cpp:163) ==31534== by 0x67D894: js::Interpret(JSContext*, js::StackFrame*, unsigned int, js::InterpMode) (MethodJIT-inl.h:75) ==31534== by 0x64835B: UncachedInlineCall(js::VMFrame&, unsigned int, void**, bool*, unsigned int) (InvokeHelpers.cpp:432) ==31534== by 0x648784: js::mjit::stubs::UncachedCallHelper(js::VMFrame&, unsigned int, js::mjit::stubs::UncachedCallResult*) (InvokeHelpers.cpp:503) ==31534== by 0x631764: CallCompiler::update() (MonoIC.cpp:1034) ==31534== by 0x62D382: js::mjit::ic::Call(js::VMFrame&, js::mjit::ic::CallICInfo*) (MonoIC.cpp:1094) ==31534== by 0x41AEC4A: ??? ==31534== by 0x5C658D: js::mjit::EnterMethodJIT(JSContext*, js::StackFrame*, void*, js::Value*) (MethodJIT.cpp:882) ==31534== by 0x5C679C: js::mjit::JaegerShotAtSafePoint(JSContext*, void*) (MethodJIT.cpp:912) ==31534== Address 0x20 is not stack'd, malloc'd or (recently) free'd ==31534== ==31534== ==31534== Process terminating with default action of signal 11 (SIGSEGV)
Comment 1•13 years ago
|
||
Hmm, the problem with -m was fixed yesterday afternoon. What does this do for you on tip?
Reporter | ||
Comment 2•13 years ago
|
||
(In reply to comment #1) > Hmm, the problem with -m was fixed yesterday afternoon. What does this do > for you on tip? You are right, now it only crashes with -m -n in the function DoMatch.
Reporter | ||
Updated•13 years ago
|
Summary: TI: Crash[@ js::mjit::Compiler::restoreVarType] or [@ DoMatch] → TI: Crash [@ DoMatch]
Comment 3•13 years ago
|
||
This WFM now, can you still reproduce? I can repro on the old changeset, but this doesn't look GC related so it is probably legitimately fixed now.
Updated•13 years ago
|
Crash Signature: [@ DoMatch]
Updated•13 years ago
|
Status: NEW → RESOLVED
Closed: 13 years ago
Resolution: --- → WORKSFORME
You need to log in
before you can comment on or make changes to this bug.
Description
•