Thu, 05 Jan 2012 17:02:58 -0800 |
Christian Wimmer |
Remove lock information from frame states, and compute it instead when LIR is generated.
|
Thu, 05 Jan 2012 17:02:13 -0800 |
Christian Wimmer |
jsr return addresses are valid in registers
|
Thu, 05 Jan 2012 16:09:47 -0800 |
Christian Wimmer |
Renaming
|
Thu, 05 Jan 2012 20:15:45 +0100 |
Gilles Duboscq |
Add SPECjvm2008 to benchmarks, use dacapos defined in sanitycheck for the gate
|
Thu, 05 Jan 2012 17:25:28 +0100 |
Gilles Duboscq |
Fix bug in code installer that caused builds with assertions to crash
|
Thu, 05 Jan 2012 14:53:37 +0100 |
Gilles Duboscq |
add a simple specjvm2008 command, fail a benchmark if one of the failureRE matches
|
Thu, 05 Jan 2012 14:25:58 +0100 |
Doug Simon |
Environment variables for HotSpot's make are only set by mx.build if they aren't already defined.
|
Thu, 05 Jan 2012 13:17:15 +0100 |
Doug Simon |
Made the gate clean the Java projects only.
|
Thu, 05 Jan 2012 11:31:46 +0100 |
Doug Simon |
Made mx.run work properly when stderr is redirected to stdout.
|
Wed, 04 Jan 2012 23:34:15 +0100 |
Thomas Wuerthinger |
Merge.
|
Wed, 04 Jan 2012 22:08:17 +0100 |
Thomas Wuerthinger |
Renaming of VMExits and VMEntries part 4.
|
Wed, 04 Jan 2012 21:13:44 +0100 |
Thomas Wuerthinger |
Renaming of VMExits and VMEntries part 3.
|
Wed, 04 Jan 2012 21:07:44 +0100 |
Thomas Wuerthinger |
Renaming of VMExits and VMEntries part 2.
|
Wed, 04 Jan 2012 20:59:11 +0100 |
Thomas Wuerthinger |
Renaming of VMExits and VMEntries part 1.
|
Wed, 04 Jan 2012 22:54:27 +0100 |
Gilles Duboscq |
Fix javac build issues with classpath
|
Wed, 04 Jan 2012 19:58:51 +0100 |
Gilles Duboscq |
Merge
|
Wed, 04 Jan 2012 19:57:51 +0100 |
Gilles Duboscq |
Various small fix around benchmarks in gate scripts
|
Wed, 04 Jan 2012 19:47:17 +0100 |
Gilles Duboscq |
fix for when no vm build is explicitely selected
|
Wed, 04 Jan 2012 13:52:46 +0100 |
Gilles Duboscq |
Fix mx : commands' scripts mx_init hook should be called before parsing command line arguments.
|
Wed, 04 Jan 2012 10:49:09 -0800 |
Christian Wimmer |
Using the incoming stack slot of object parameters is broken, so disable it for now.
|
Tue, 03 Jan 2012 18:47:27 -0800 |
Christian Wimmer |
Do not emit stack-to-stack moves.
|
Tue, 03 Jan 2012 18:42:32 -0800 |
Christian Wimmer |
Define incoming parameters in a PARAMS instruction. This simplifies register allocation because the initial register state does not have to be inferred from the calling convention.
|
Tue, 03 Jan 2012 18:41:01 -0800 |
Christian Wimmer |
Do not emit stack-to-stack moves.
|
Tue, 03 Jan 2012 18:36:53 -0800 |
Christian Wimmer |
Improve debug printing
|
Tue, 03 Jan 2012 18:35:37 -0800 |
Christian Wimmer |
Remove stack-to-stack moves because 32-bit memory push and pop instructions are not available on Intel 64 bit.
|
Tue, 03 Jan 2012 18:33:54 -0800 |
Christian Wimmer |
Improve debug printing
|
Tue, 03 Jan 2012 13:47:14 -0800 |
Christian Wimmer |
Allow an arbitary number of output operands for LIR instructions
|
Tue, 03 Jan 2012 12:10:27 -0800 |
Christian Wimmer |
Remove CiVariable from the CRI
|
Tue, 03 Jan 2012 18:22:10 +0100 |
Thomas Wuerthinger |
moved graphbuilder into the graal.java project.
|
Tue, 03 Jan 2012 17:53:26 +0100 |
Thomas Wuerthinger |
Fixed unit tests.
|
Tue, 03 Jan 2012 17:31:23 +0100 |
Thomas Wuerthinger |
Add graph builder to the phase plan.
|
Tue, 03 Jan 2012 16:47:54 +0100 |
Thomas Wuerthinger |
Canonicalized projects.
|
Tue, 03 Jan 2012 16:47:02 +0100 |
Thomas Wuerthinger |
Merge.
|
Tue, 03 Jan 2012 16:44:31 +0100 |
Thomas Wuerthinger |
Move printer into separate project.
|
Tue, 03 Jan 2012 16:29:28 +0100 |
Thomas Wuerthinger |
Renamed cri packages from sun to oracle.
|
Tue, 03 Jan 2012 16:24:22 +0100 |
Thomas Wuerthinger |
More restructuring of cri; moved bytecodes ingo compiler.graphbuilder.* package
|
Tue, 03 Jan 2012 16:06:31 +0100 |
Thomas Wuerthinger |
Move HotSpotMethod* and HotSpotType* implementations into ri subpackage.
|
Tue, 03 Jan 2012 15:36:28 +0100 |
Thomas Wuerthinger |
Remove isFoldable from runtime interface.
|
Tue, 03 Jan 2012 15:34:46 +0100 |
Thomas Wuerthinger |
Removed inlining decisions from CRI.
|
Tue, 03 Jan 2012 15:30:57 +0100 |
Thomas Wuerthinger |
More CRI clean up.
|
Tue, 03 Jan 2012 15:26:18 +0100 |
Thomas Wuerthinger |
Delete HotSpotTarget.
|
Tue, 03 Jan 2012 15:24:41 +0100 |
Thomas Wuerthinger |
CRI clean up.
|
Tue, 03 Jan 2012 15:17:35 +0100 |
Thomas Wuerthinger |
Removed tail call node and opcode (=> must be moved to the truffle project).
|
Tue, 03 Jan 2012 15:16:05 +0100 |
Thomas Wuerthinger |
Added mx.bat for Windows.
|
Tue, 03 Jan 2012 15:15:53 +0100 |
Thomas Wuerthinger |
Fixed warning.
|
Tue, 03 Jan 2012 15:49:22 +0100 |
Doug Simon |
Downloads are attempted via Java first since it shows a progress counter.
|
Tue, 03 Jan 2012 13:54:11 +0100 |
Doug Simon |
Disabled copyright check in the gate until the CheckCopyright tools supports the HotSpot copyright notice format.
|
Tue, 03 Jan 2012 13:53:38 +0100 |
Doug Simon |
Made stdout for mx unbuffered.
|
Tue, 03 Jan 2012 12:56:55 +0100 |
Doug Simon |
Added message showing the number of changesets being checked.
|
Tue, 03 Jan 2012 12:37:31 +0100 |
Doug Simon |
Added copyright check to the gate.
|
Mon, 02 Jan 2012 17:39:20 -0800 |
Christian Wimmer |
Move all isXxx and asXxx out of CiValue and into their own util class.
|
Mon, 02 Jan 2012 14:38:17 -0800 |
Christian Wimmer |
Simplify CiValue
|
Mon, 02 Jan 2012 14:16:08 -0800 |
Christian Wimmer |
Simplify FrameMap: make offsets of spill slots and outgoing parameters independent so that they can be allocated at the same time, eliminating the separate phases. This makes the separate StackBlock unnecesary. Change CiStackSlot to use byte offsets instead of spill slot index. This makes CiTarget.spillSlotSize unnecessary.
|
Mon, 02 Jan 2012 22:18:16 +0100 |
Thomas Wuerthinger |
Fixed gate script such that JAVA_HOME can be set in env file.
|
Mon, 02 Jan 2012 21:55:47 +0100 |
Thomas Wuerthinger |
Merge.
|
Mon, 02 Jan 2012 21:52:23 +0100 |
Thomas Wuerthinger |
Add the Dacapo benchmarks to the "lib" folder instead of using the environment variable. Fixed an issue in the downloader.
|