Mon, 01 Jun 2015 14:17:05 +0200 |
Stefan Anzinger |
JVMCI update makefiles for Solaris/SPARC; os::read at early stage of VM does not work, use ::read as fallback in JVMCIRuntime::parse_lines (JBS:GRAAL-54)
|
Mon, 01 Jun 2015 13:19:48 +0200 |
Thomas Wuerthinger |
Merge.
|
Mon, 01 Jun 2015 00:44:38 +0200 |
Thomas Wuerthinger |
Merge.
|
Sun, 31 May 2015 23:58:09 +0200 |
Thomas Wuerthinger |
Merge.
|
Sun, 31 May 2015 23:57:57 +0200 |
Thomas Wuerthinger |
Enable inlining during parsing by default.
|
Sun, 31 May 2015 23:20:24 +0200 |
Thomas Wuerthinger |
Fix in the bytecode parser: Do not overwrite the stateAfter of the start instruction.
|
Sun, 31 May 2015 23:05:25 +0200 |
Thomas Wuerthinger |
Small fix to NestedLoopTest: Make sure the placeholder methods cannot be inlined.
|
Sun, 31 May 2015 22:48:50 +0200 |
Thomas Wuerthinger |
Fix in the dominator based conditional elimination for the corner case of a loop exit merge.
|
Sun, 31 May 2015 13:27:50 +0200 |
Thomas Wuerthinger |
Fix for the method handle plugin.
|
Mon, 01 Jun 2015 12:29:48 +0200 |
Doug Simon |
don't destroy and clear JNI global for HotSpotJVMCIRuntime singleton object during shutdown as racing compiler threads may still access it
|
Sun, 31 May 2015 23:57:31 +0200 |
Doug Simon |
recast VM compilation request via JVMCI to be a HotSpot JVMCI event, removing the last direct reference to a Graal class in the VM (JBS:GRAAL-53)
|
Sun, 31 May 2015 14:30:13 +0200 |
Doug Simon |
handle service filtering that returns an empty result
|
Sun, 31 May 2015 13:42:47 +0200 |
Doug Simon |
Truffle and NFI implementations are now accessed via JVMCI services instead of being hard coded in the VM (JBS:GRAAL-51)
|
Sun, 31 May 2015 12:32:15 +0200 |
Doug Simon |
replaced use of javap with a small Java utility for refining service class list to only those implementing com.oracle.jvmci.service.Service
|