Mercurial > hg > graal-compiler
graph
-
don't synthesize a deoptState in ForeignCallNode is canDeoptimize == falseWed, 05 Jun 2013 14:18:35 +0200, by Lukas Stadler
-
HotSpotRuntime should decide when to lower which nodes, not the nodes themselvesWed, 05 Jun 2013 13:10:42 +0200, by Lukas Stadler
-
allow for late lowering of MemoryCheckpoints (handle usages by FloatingReads in SnippetTemplate)Wed, 05 Jun 2013 12:05:27 +0200, by Lukas Stadler
-
simplify FrameStateAssignmentPhase, add guarantee that every DeoptimizingNode has a FrameStateWed, 05 Jun 2013 11:50:07 +0200, by Lukas Stadler
-
PEA: changes to allow BlockState to be extendedWed, 05 Jun 2013 11:47:46 +0200, by Lukas Stadler
-
PEA phase only needs PhaseContextWed, 05 Jun 2013 11:46:19 +0200, by Lukas Stadler
-
make ReadNode and WriteNode virtualizableWed, 05 Jun 2013 11:45:13 +0200, by Lukas Stadler
-
only create overflow guards for loops that have safepointsWed, 05 Jun 2013 11:44:53 +0200, by Lukas Stadler
-
use loadHub without null check in MonitorSnippetsWed, 05 Jun 2013 11:44:05 +0200, by Lukas Stadler
-
use StubForeignCallNode within stubs, instead of ForeignCallNodeWed, 05 Jun 2013 11:34:55 +0200, by Lukas Stadler
-
LIRGenerator.emitForeignCall uses linkage to determine if a state is neededWed, 05 Jun 2013 11:10:08 +0200, by Lukas Stadler
-
pull HotSpotForeignCallLinkage.isLeaf into ForeignCallLinkage and rename to canDeoptimizeWed, 05 Jun 2013 11:09:33 +0200, by Lukas Stadler
-
TestResolvedJavaMethod: relax check for max stack sizeWed, 05 Jun 2013 11:48:53 +0200, by Bernhard Urban
-
VerifyUsageWithEquals: fix wordingFri, 31 May 2013 11:38:35 +0200, by Bernhard Urban