Tue, 17 Jul 2012 11:55:57 +0200 |
Doug Simon |
moved HotSpotXirGenerator to com.oracle.graal.hotspot.target.amd64
|
Tue, 17 Jul 2012 23:57:18 +0200 |
Gilles Duboscq |
Make ReadEliminationPhase Value proxy aware now that FLoatingRead keeps loop closed form
|
Tue, 17 Jul 2012 20:07:53 +0200 |
Gilles Duboscq |
Merge
|
Tue, 17 Jul 2012 20:07:00 +0200 |
Gilles Duboscq |
Make FloatingRead phase respect loop closed form and use PostOrderNodeIterator
|
Tue, 17 Jul 2012 11:39:50 +0200 |
Doug Simon |
fixed bugs related to -G:+InlineVTableStubs and re-enabled it by default
|
Mon, 16 Jul 2012 22:09:21 +0200 |
Doug Simon |
disabled vtable inline dispatch by default until remaining issues are resolved
|
Mon, 16 Jul 2012 15:24:06 +0200 |
Doug Simon |
Merge.
|
Mon, 16 Jul 2012 15:20:50 +0200 |
Doug Simon |
vtable dispatch inlining for megamorphic virtual calls now works and is enabled by default
|
Mon, 16 Jul 2012 15:19:25 +0200 |
Doug Simon |
parsing and patching of variable sized NativeCallReg instructions now works properly
|
Mon, 16 Jul 2012 15:18:38 +0200 |
Doug Simon |
renaming in native code for CiTargetMethod -> CompilationResult
|
Mon, 16 Jul 2012 11:07:07 +0200 |
Gilles Duboscq |
Merge
|
Mon, 16 Jul 2012 11:04:32 +0200 |
Gilles Duboscq |
Fix ConcurrentModificationException in IGV
|
Fri, 13 Jul 2012 17:43:11 +0200 |
Gilles Duboscq |
Disable read canoncialization until we have a viable solution for non-object oops
|
Fri, 13 Jul 2012 15:51:38 +0200 |
Gilles Duboscq |
Additional fix in read canonicalization
|
Fri, 13 Jul 2012 15:41:01 +0200 |
Gilles Duboscq |
Remove useless double canonicalization
|
Fri, 13 Jul 2012 15:33:02 +0200 |
Gilles Duboscq |
Fix canonicalisation of final reads and factor out read/floating read canonicalization code
|
Fri, 13 Jul 2012 15:31:53 +0200 |
Gilles Duboscq |
Revert mistakenly commited FloatingRead phase changes
|
Fri, 13 Jul 2012 14:10:02 +0200 |
Gilles Duboscq |
Add a alwaysNull property to ObjectStamp (Phi(null, a! A) get a "a A" stamp and not just "a -")
|
Fri, 13 Jul 2012 11:29:15 +0200 |
Gilles Duboscq |
Do without the Top stamp for now, too little benefits
|
Sat, 14 Jul 2012 21:38:19 +0200 |
Doug Simon |
incomplete (non XIR) support for inlining virtual dispatch at call sites - all design questions yet to be addressed
|
Sat, 14 Jul 2012 21:28:23 +0200 |
Doug Simon |
added pc offset within nmethod to output for TraceExceptions
|
Sat, 14 Jul 2012 21:26:14 +0200 |
Doug Simon |
Removed alignment for register indirect calls
|
Fri, 13 Jul 2012 14:52:35 +0200 |
Doug Simon |
fixed bug in node for calling the verify oop stub
|
Fri, 13 Jul 2012 11:40:41 +0200 |
Doug Simon |
Merge.
|
Fri, 13 Jul 2012 09:06:14 +0200 |
Doug Simon |
mx now handles unambiguous command prefixes (e.g. instead of 'mx eclipseinit' you can use 'mx ec')
|
Thu, 12 Jul 2012 17:16:34 +0200 |
Doug Simon |
partial (non XIR) support for inlining virtual dispatch at call sites - still needs fixing
|
Thu, 12 Jul 2012 12:04:27 +0200 |
Doug Simon |
moved HIR lowering of Invokes into HotspotRuntime
|
Thu, 12 Jul 2012 11:55:03 +0200 |
Doug Simon |
added support to mx for ignoring (broken) projects
|
Thu, 12 Jul 2012 18:58:36 +0200 |
Gilles Duboscq |
Make ReadEliminationPhase support phis (eliminates read when the last access is a memeory phi of writes, recursively)
|
Thu, 12 Jul 2012 16:59:09 +0200 |
Gilles Duboscq |
Make DebugDumpHandler closable, close them on compiler thread when compiler threads finish
|
Wed, 11 Jul 2012 15:38:28 +0200 |
Gilles Duboscq |
in tryCanonicalize return node.isDeleted() in the default path to be sure to catch node deletion in simplify and clean unused inputs
|
Wed, 11 Jul 2012 15:28:28 +0200 |
Gilles Duboscq |
Make sure the right port is used for dumping
|
Wed, 11 Jul 2012 14:41:22 +0200 |
Gilles Duboscq |
Merge
|
Wed, 11 Jul 2012 14:36:32 +0200 |
Gilles Duboscq |
Put Binary and xml dumpers behing a common interface and use the same dumphandler for them
|
Wed, 11 Jul 2012 14:13:38 +0200 |
Gilles Duboscq |
Adapt the coloring filter for binary dumped graphs
|
Tue, 10 Jul 2012 15:02:41 +0200 |
Gilles Duboscq |
Add ability to give an existing to getDebugProperties
|
Mon, 09 Jul 2012 15:30:20 +0200 |
Gilles Duboscq |
Remove useless Edge class
|
Mon, 09 Jul 2012 14:22:22 +0200 |
Gilles Duboscq |
Merge
|
Mon, 09 Jul 2012 14:15:55 +0200 |
Gilles Duboscq |
Add binary graph dumping to compiler (-G:+PrintBinaryGraphs)
|
Wed, 04 Jul 2012 15:27:28 +0200 |
Gilles Duboscq |
Merge
|
Wed, 04 Jul 2012 15:25:40 +0200 |
Gilles Duboscq |
Slightly more sensible work order in Canonicalizer (not unused? -> not canonicalized? -> stamp) and be sure to kill unused
|
Wed, 04 Jul 2012 15:23:51 +0200 |
Gilles Duboscq |
Small cleanups and add snapshotTo method to node iterables
|
Tue, 03 Jul 2012 16:11:04 +0200 |
Gilles Duboscq |
Kill unused floating nodes in Canonicalizer
|
Tue, 10 Jul 2012 14:11:40 +0200 |
Doug Simon |
consolidated 2 separate size fields in TLABAllocate into 1
|
Tue, 10 Jul 2012 13:27:39 +0200 |
Doug Simon |
disabled XIR lowering of invoke by default
|
Tue, 10 Jul 2012 13:22:54 +0200 |
Doug Simon |
Merge.
|
Tue, 10 Jul 2012 10:36:08 +0200 |
Doug Simon |
added null check for receiver of non-static invokes
|
Tue, 10 Jul 2012 10:08:49 +0200 |
Doug Simon |
added tests for null receiver for non-static invokes
|
Tue, 10 Jul 2012 09:36:34 +0200 |
Doug Simon |
handles changes in IGV bytecode format
|
Tue, 10 Jul 2012 10:49:31 +0200 |
Lukas Stadler |
less aggressive tail duplication
|
Mon, 09 Jul 2012 22:18:49 +0200 |
Doug Simon |
implemented non-XIR lowering of invokes (todo: inline virtual dispatch and null checking of receivers)
|
Mon, 09 Jul 2012 22:17:00 +0200 |
Doug Simon |
formatting fix
|
Mon, 09 Jul 2012 22:16:36 +0200 |
Doug Simon |
clarified error message when Value fields of LIRInstruction classes have the wrong modifiers
|
Mon, 09 Jul 2012 22:15:15 +0200 |
Doug Simon |
removed old, commented out code
|
Mon, 09 Jul 2012 22:12:12 +0200 |
Doug Simon |
use more efficient Unsafe API for getting array base and index values
|
Mon, 09 Jul 2012 22:09:36 +0200 |
Doug Simon |
added prefix to formatted object Constants to denote the type of the constant
|