Title: [185099] trunk/Source/_javascript_Core
- Revision
- 185099
- Author
- [email protected]
- Date
- 2015-06-01 19:59:39 -0700 (Mon, 01 Jun 2015)
Log Message
If a call has ever taken the virtual slow path, make sure that the DFG knows this
https://bugs.webkit.org/show_bug.cgi?id=145501
Reviewed by Geoffrey Garen.
Now now return higher fidelity information in the case of no polymorphic call stub. If the
virtual slow path was ever taken, we note this, and we note either zero or one call variant
based on the IC's last callee.
* bytecode/CallLinkStatus.cpp:
(JSC::CallLinkStatus::computeFromCallLinkInfo):
(JSC::CallLinkStatus::computeFor):
Modified Paths
Diff
Modified: trunk/Source/_javascript_Core/ChangeLog (185098 => 185099)
--- trunk/Source/_javascript_Core/ChangeLog 2015-06-02 02:55:36 UTC (rev 185098)
+++ trunk/Source/_javascript_Core/ChangeLog 2015-06-02 02:59:39 UTC (rev 185099)
@@ -1,3 +1,18 @@
+2015-05-31 Filip Pizlo <[email protected]>
+
+ If a call has ever taken the virtual slow path, make sure that the DFG knows this
+ https://bugs.webkit.org/show_bug.cgi?id=145501
+
+ Reviewed by Geoffrey Garen.
+
+ Now now return higher fidelity information in the case of no polymorphic call stub. If the
+ virtual slow path was ever taken, we note this, and we note either zero or one call variant
+ based on the IC's last callee.
+
+ * bytecode/CallLinkStatus.cpp:
+ (JSC::CallLinkStatus::computeFromCallLinkInfo):
+ (JSC::CallLinkStatus::computeFor):
+
2015-06-01 Michael Saboff <[email protected]>
Crash in com.apple.WebKit.WebContent at com.apple._javascript_Core: JSC::revertCall + 24
Modified: trunk/Source/_javascript_Core/bytecode/CallLinkStatus.cpp (185098 => 185099)
--- trunk/Source/_javascript_Core/bytecode/CallLinkStatus.cpp 2015-06-02 02:55:36 UTC (rev 185098)
+++ trunk/Source/_javascript_Core/bytecode/CallLinkStatus.cpp 2015-06-02 02:59:39 UTC (rev 185099)
@@ -150,6 +150,11 @@
// that is still marginally valid (i.e. the pointers ain't stale). This kind of raciness
// is probably OK for now.
+ // FIXME: If the GC often clears this call, we should probably treat it like it always takes the
+ // slow path. We could be smart about this; for example if we cleared a specific callee but the
+ // despecified executable was alive then we could note that separately.
+ // https://bugs.webkit.org/show_bug.cgi?id=145502
+
// PolymorphicCallStubRoutine is a GCAwareJITStubRoutine, so if non-null, it will stay alive
// until next GC even if the CallLinkInfo is concurrently cleared. Also, the variants list is
// never mutated after the PolymorphicCallStubRoutine is instantiated. We have some conservative
@@ -209,17 +214,18 @@
return result;
}
- if (callLinkInfo.slowPathCount >= Options::couldTakeSlowCaseMinimumCount())
- return takesSlowPath();
+ CallLinkStatus result;
- JSFunction* target = callLinkInfo.lastSeenCallee.get();
- if (!target)
- return takesSlowPath();
+ if (JSFunction* target = callLinkInfo.lastSeenCallee.get()) {
+ CallVariant variant(target);
+ if (callLinkInfo.hasSeenClosure)
+ variant = variant.despecifiedClosure();
+ result.m_variants.append(variant);
+ }
- if (callLinkInfo.hasSeenClosure)
- return CallLinkStatus(target->executable());
+ result.m_couldTakeSlowPath = !!callLinkInfo.slowPathCount;
- return CallLinkStatus(target);
+ return result;
}
CallLinkStatus CallLinkStatus::computeFor(
_______________________________________________
webkit-changes mailing list
[email protected]
https://lists.webkit.org/mailman/listinfo/webkit-changes
