Changeset 155480 in webkit for trunk/Source/JavaScriptCore/dfg/DFGUseKind.h

Timestamp:

Sep 10, 2013, 2:55:45 PM (12 years ago)

Author:

[email protected]

Message:

Introduce a SpecInt48 type and be more careful about what we mean by "Top"
​https://bugs.webkit.org/show_bug.cgi?id=121116

Reviewed by Oliver Hunt.

SpecInt48 will mean that we have something that would be a double if it was a JSValue,
but it's profitable to represent it as something other than a double.

SpecInt48AsDouble means that it has a value that could have been represented like
SpecInt48, but we're making a heuristic decision not to do it.

• bytecode/SpeculatedType.h:

(JSC::isInt48Speculation):

• dfg/DFGAbstractInterpreterInlines.h:

(JSC::DFG::::executeEffects):
(JSC::DFG::::clobberCapturedVars):

• dfg/DFGAbstractValue.cpp:

(JSC::DFG::AbstractValue::filter):

• dfg/DFGAbstractValue.h:

(JSC::DFG::AbstractValue::makeHeapTop):
(JSC::DFG::AbstractValue::makeBytecodeTop):
(JSC::DFG::AbstractValue::isHeapTop):
(JSC::DFG::AbstractValue::heapTop):
(JSC::DFG::AbstractValue::validateType):
(JSC::DFG::AbstractValue::validate):
(JSC::DFG::AbstractValue::makeTop):

• dfg/DFGInPlaceAbstractState.cpp:

(JSC::DFG::InPlaceAbstractState::initialize):

• dfg/DFGJITCompiler.h:

(JSC::DFG::JITCompiler::noticeOSREntry):

• dfg/DFGUseKind.h:

(JSC::DFG::typeFilterFor):

File:

• trunk/Source/JavaScriptCore/dfg/DFGUseKind.h (modified) (2 diffs)

trunk/Source/JavaScriptCore/dfg/DFGUseKind.h

@@ -63 +63 @@
     switch (useKind) {
     case UntypedUse:
-        return SpecEmptyOrTop; // TOP isn't good enough; untyped uses may use the normally unseen empty value, in the case of lazy registers.
+        return SpecFullTop;
     case Int32Use:
     case KnownInt32Use:
@@ -98 +98 @@
     default:
         RELEASE_ASSERT_NOT_REACHED();
-        return SpecTop;
+        return SpecFullTop;
     }
 }