Title: [199508] trunk/Source/_javascript_Core
- Revision
- 199508
- Author
- fpi...@apple.com
- Date
- 2016-04-13 12:04:32 -0700 (Wed, 13 Apr 2016)
Log Message
PolymorphicAccess::regenerate() shouldn't have to clone non-generated AccessCases
https://bugs.webkit.org/show_bug.cgi?id=156493
Reviewed by Geoffrey Garen.
Cloning AccessCases is only necessary if they hold some artifacts that are used by code that
they already generated. So, if the state is not Generated, we don't have to bother with
cloning them.
This should speed up PolymorphicAccess regeneration a bit more.
* bytecode/PolymorphicAccess.cpp:
(JSC::AccessCase::commit):
(JSC::PolymorphicAccess::regenerate):
Modified Paths
Diff
Modified: trunk/Source/_javascript_Core/ChangeLog (199507 => 199508)
--- trunk/Source/_javascript_Core/ChangeLog 2016-04-13 18:52:48 UTC (rev 199507)
+++ trunk/Source/_javascript_Core/ChangeLog 2016-04-13 19:04:32 UTC (rev 199508)
@@ -1,3 +1,20 @@
+2016-04-12 Filip Pizlo <fpi...@apple.com>
+
+ PolymorphicAccess::regenerate() shouldn't have to clone non-generated AccessCases
+ https://bugs.webkit.org/show_bug.cgi?id=156493
+
+ Reviewed by Geoffrey Garen.
+
+ Cloning AccessCases is only necessary if they hold some artifacts that are used by code that
+ they already generated. So, if the state is not Generated, we don't have to bother with
+ cloning them.
+
+ This should speed up PolymorphicAccess regeneration a bit more.
+
+ * bytecode/PolymorphicAccess.cpp:
+ (JSC::AccessCase::commit):
+ (JSC::PolymorphicAccess::regenerate):
+
2016-04-13 Mark Lam <mark....@apple.com>
ES6: Implement String.prototype.split and RegExp.prototype[@@split].
Modified: trunk/Source/_javascript_Core/bytecode/PolymorphicAccess.cpp (199507 => 199508)
--- trunk/Source/_javascript_Core/bytecode/PolymorphicAccess.cpp 2016-04-13 18:52:48 UTC (rev 199507)
+++ trunk/Source/_javascript_Core/bytecode/PolymorphicAccess.cpp 2016-04-13 19:04:32 UTC (rev 199508)
@@ -399,7 +399,10 @@
Vector<WatchpointSet*, 2> AccessCase::commit(VM& vm, const Identifier& ident)
{
- RELEASE_ASSERT(m_state == Primordial);
+ // It's fine to commit something that is already committed. That arises when we switch to using
+ // newly allocated watchpoints. When it happens, it's not efficient - but we think that's OK
+ // because most AccessCases have no extra watchpoints anyway.
+ RELEASE_ASSERT(m_state == Primordial || m_state == Committed);
Vector<WatchpointSet*, 2> result;
@@ -1546,28 +1549,36 @@
// to be unmutated. For sure, we want it to hang onto any data structures that may be referenced
// from the code of the current stub (aka previous).
ListType cases;
- for (unsigned i = 0; i < m_list.size(); ++i) {
- AccessCase& someCase = *m_list[i];
- // Ignore cases that cannot possibly succeed anymore.
- if (!someCase.couldStillSucceed())
- continue;
+ unsigned srcIndex = 0;
+ unsigned dstIndex = 0;
+ while (srcIndex < m_list.size()) {
+ std::unique_ptr<AccessCase> someCase = WTFMove(m_list[srcIndex++]);
- // Figure out if this is replaced by any later case.
- bool found = false;
- for (unsigned j = i + 1; j < m_list.size(); ++j) {
- if (m_list[j]->canReplace(someCase)) {
- found = true;
- break;
+ // If the case had been generated, then we have to keep the original in m_list in case we
+ // fail to regenerate. That case may have data structures that are used by the code that it
+ // had generated. If the case had not been generated, then we want to remove it from m_list.
+ bool isGenerated = someCase->state() == AccessCase::Generated;
+
+ [&] () {
+ if (!someCase->couldStillSucceed())
+ return;
+
+ // Figure out if this is replaced by any later case.
+ for (unsigned j = srcIndex; j < m_list.size(); ++j) {
+ if (m_list[j]->canReplace(*someCase))
+ return;
}
- }
- if (found)
- continue;
+
+ if (isGenerated)
+ cases.append(someCase->clone());
+ else
+ cases.append(WTFMove(someCase));
+ }();
- // FIXME: Do we have to clone cases that aren't generated? Maybe we can just take those
- // from m_list, since we don't have to keep those alive if we fail.
- // https://bugs.webkit.org/show_bug.cgi?id=156493
- cases.append(someCase.clone());
+ if (isGenerated)
+ m_list[dstIndex++] = WTFMove(someCase);
}
+ m_list.resize(dstIndex);
if (verbose)
dataLog("In regenerate: cases: ", listDump(cases), "\n");
_______________________________________________
webkit-changes mailing list
webkit-changes@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-changes
