Mercurial > hg > truffle
graph
-
Removed HandleException runtime call and associated C++ code (unused).Sat, 25 Feb 2012 00:52:58 +0100, by Thomas Wuerthinger
-
Simplified and improved the speed of compiled exception handler lookup based on the Graal exception handler model.Sat, 25 Feb 2012 00:49:04 +0100, by Thomas Wuerthinger
-
Simplify exception debug information and exception handler table creation based on Graal's simplified model of exception handler information (i.e. the dispatch is done in compiled code).Sat, 25 Feb 2012 00:40:34 +0100, by Thomas Wuerthinger
-
Fixed a wrong frame state for the exception edge after a polymorphic inlining. Removed a NPE allocation in a runtime call branch introduced for Graal.Fri, 24 Feb 2012 22:32:43 +0100, by Thomas Wuerthinger
-
Some DCE in codeInstaller/exceptions.Fri, 24 Feb 2012 20:39:12 +0100, by Thomas Wuerthinger
-
Merge.Fri, 24 Feb 2012 20:05:32 +0100, by Thomas Wuerthinger
-
Make sure rethrowException flag is cleared also when return/unwind blocks are entered, they could use the framestate for the synchronization epilogue.Fri, 24 Feb 2012 19:52:27 +0100, by Thomas Wuerthinger
-
Made DeoptimizeALot flag available in product builds (to make testing with that flag faster).Fri, 24 Feb 2012 19:51:18 +0100, by Thomas Wuerthinger
-
Removed dead code from the inlining util. STATE_BEFORE is only used in BlockBegin, which in turn is not added to the caller graph during inlining.Fri, 24 Feb 2012 19:50:23 +0100, by Thomas Wuerthinger
-
MergeFri, 24 Feb 2012 17:17:48 +0100, by Gilles Duboscq
-
MergeFri, 24 Feb 2012 17:16:39 +0100, by Gilles Duboscq
-
Activate jtt for the gateFri, 24 Feb 2012 17:16:03 +0100, by Gilles Duboscq
-
add hprofs to .hgignoreWed, 22 Feb 2012 13:19:32 +0100, by Gilles Duboscq
-
Better ConvertDeoptimizeToGuardPhase (experimental).Fri, 24 Feb 2012 13:42:24 +0100, by Andreas Woess
-
added bytecode name to deoptimization detailsThu, 23 Feb 2012 16:57:05 -0800, by Christian Haeubl
-
Fixed two asserts in the implicit div exception handling for Windows that are not valid for Graal (as it deoptimizes to some place before the div instead of exactly to the div bytecode).Thu, 23 Feb 2012 23:06:28 +0100, by Thomas Wuerthinger
-
Remove the recording of oop arguments that are passed on the stack.Thu, 23 Feb 2012 22:04:35 +0100, by Thomas Wuerthinger
-
Fix reexecute boolean in HotSpot debug information. Introduce "duringCall" flag in FrameState that indicates that the bci of the frame state denotes an invoke that should *not* be reexecuted.Thu, 23 Feb 2012 21:43:59 +0100, by Thomas Wuerthinger
-
Merge.Thu, 23 Feb 2012 12:06:39 +0100, by Thomas Wuerthinger
-
Merge.Wed, 22 Feb 2012 21:24:08 +0100, by Thomas Wuerthinger
-
Small changes to the deopt stub.Wed, 22 Feb 2012 14:07:10 +0100, by Thomas Wuerthinger
-
changed TraceDeoptimization outputWed, 22 Feb 2012 18:20:11 -0800, by Christian Haeubl
-
another fix for resolving the issue with biased monitor and deoptimizationWed, 22 Feb 2012 16:53:23 -0800, by Christian Haeubl
-
fix NPE in tests caused by recordMethodContentsWed, 22 Feb 2012 17:27:11 +0100, by Lukas Stadler
-
add Condition.join, Condition.meet and testsWed, 22 Feb 2012 17:05:20 +0100, by Lukas Stadler
-
add MethodContents assumption to fix debbugging issueWed, 22 Feb 2012 17:04:27 +0100, by Lukas Stadler
-
bugfixTue, 21 Feb 2012 16:23:44 -0800, by Christian Haeubl
-
added flag PrintDeoptimizationDetailsTue, 21 Feb 2012 16:05:46 -0800, by Christian Haeubl