May 2015 - aarch64-port-dev

Support large codecache for JDK8
by Andrew Dinn 26 Nov '15

26 Nov '15

Attached for review is a patch which adds large codecache support to jdk8 (jdk7 version will be next). It passes basic smoke tests (java hello, javac Hello.java, netbeans build and run simple project). Ok to push? regards, Andrew Dinn -----------

3 7

RFR: 8079565: aarch64: Add vectorization support for aarch64
by Edward Nevill 04 Jun '15

04 Jun '15

Hi, The following webrev http://cr.openjdk.java.net/~enevill/8079565/webrev.00/ adds support for vectorization on aarch64. This is an initial pass at adding vectorization. There are a number of limitations. - Only 128 bit vectors are supported. The current implementation only supports vectors which are exactly 128 bits in length. Support needs to be added for shorter vectors (64 / 32??). - The pack/unpack vectorizations are missing - The Replicate opcode is suboptimal. Currently it just uses a sequence of MOVI/ORRI or MVNI/BICI instructions to replicate an immediate value across a vector. This can take up to 4 instructions to replicate a 32 bit value across the vector (1 MOVI and 3 ORRIs ot 1 MVNI and 3 BICIs). - The cost model needs tuning. At the moment most vectorizations are just costed at 1 X instruction cost. - It needs benchmarking and tuning across different partners hardware. For example, on some partners hardware it may not be worthwhile performing the Long or Double vectorizations. I have done some performance testing on one partners hardware using the hotspot vector tests with the following results:- Byte Vectors: approx 3-4 X improvement Short Vectors: approx 2.5-3.25 X improvement Int Vectors: approx 1.5-2.5 X improvement Long Vectors: approx 1.0-1.33 X improvement Float Vectors: approx 1.4-1.8 X improvement Double Vectors: approx 0.85-1.25 X improvement I have also implemented the scalar reduction optimization with the following results:- Scalar Sum Reduction Int: ~4.6 X improvement Scalar Sum Reduction Float: ~2.3 X improvement Scalar Sum Reduction Double: ~1.9 X improvement Scalar Product Reduction Int: ~1.2 X improvement Scalar Product Reduction Float: ~1.1 X improvement Scalar Product Reduction Double: ~0.8 X improvement Tested with JTreg hotspot Original: Test results: passed: 814; failed: 32; error: 3 Revised : Test results: passed: 814; failed: 32; error: 3 Langtools: Test results: passed: 3,222; error: 11 Please review, Thanks, Ed.

3 4

RFR: 8081289: aarch64: add support for RewriteFrequentPairs in interpreter
by Edward Nevill 04 Jun '15

04 Jun '15

Hi, The following webrev adds support for RewriteFrequentPairs to the template interpreter for aarch64. http://cr.openjdk.java.net/~enevill/8081289/webrev.00 This was contributed by Alexander Alexeev (alexander.alexeev(a)caviumnetworks.com) This gives a small improvement to the interpreter on aarch64, and brings it in line with all the other ports (x86, sparc, ppc, zero) which all support RewriteFrequentPairs. I have done some performance measurement using -Xint with some micro benchmarks and I see a small improvement on each. java dhrystone: +9% embedded caffeinemark: +4% grinderbench: +1% dacapo (avrora): +1% Tested with hotspot jtreg:- Original: Test results: passed: 787; failed: 24; error: 44 With patch: Test results: passed: 785; failed: 24; error: 46 The difference in the # of errors is due to timeouts because we are running -Xint. Please review and if OK I will push. All the best, Ed.

2 2

RFR: Backports from jdk9 to jdk8
by Edward Nevill 29 May '15

29 May '15

Hi, The following webrev backports a number of changesets from jdk9 to jdk8. http://cr.openjdk.java.net/~enevill/jdk8-backports/ The changesets are backported are listed below. Where possible I have kept the changeset unchanged by export from jdk9 and import direct into jdk8. In the case of some changesets this was not possible because of conflicts. In these cases I have applied the changes manually. However in all cases I have attributed the author of the changeset correctly. I have tested with jtreg hotspot with the following results:- Original: Test results: passed: 650; failed: 19 Patched: Test results: passed: 650; failed: 19 OK to push to jdk8? Ed. --- CUT HERE --- changeset: 8156:d38b6415fcd8 tag: tip user: enevill date: Wed May 27 15:40:40 2015 +0100 files: src/cpu/aarch64/vm/stubGenerator_aarch64.cpp description: 8080586: aarch64: hotspot test compiler/codegen/7184394/TestAESMain.java fails Summary: Return correct length in generate_cipherBlockChaining_encryptAESCrypt changeset: 8155:adcffd0e1707 user: enevill date: Wed May 27 15:28:40 2015 +0100 files: src/cpu/aarch64/vm/aarch64.ad src/cpu/aarch64/vm/vm_version_aarch64.cpp src/cpu/aarch64/vm/vm_version_aarch64.hpp description: 8079203: AARCH64: Need to cater for different partner implementations Summary: Parse /proc/cpuinfo to derive implementation specific info changeset: 8154:e84a2db0758d user: aph date: Tue Apr 14 17:19:08 2015 +0100 files: src/cpu/aarch64/vm/interp_masm_aarch64.hpp src/cpu/aarch64/vm/register_aarch64.hpp description: 8076467: AARCH64: assertion fail with -XX:+UseG1GC Summary: Don't call encoding unless bool is true. Reviewed-by: kvn changeset: 8153:394a87600c41 user: enevill date: Fri Apr 24 11:01:37 2015 +0000 files: src/cpu/aarch64/vm/aarch64.ad src/cpu/aarch64/vm/frame_aarch64.inline.hpp description: 8075930: AARCH64: Use FP Register in C2 Summary: modify to allow C2 to allocate FP (R29) as a general register Reviewed-by: aph, kvn, dlong changeset: 8152:5d3f35c13442 user: aph date: Fri Mar 20 17:39:29 2015 +0000 files: src/cpu/aarch64/vm/aarch64.ad src/cpu/aarch64/vm/aarch64_ad.m4 description: 8075443: AARCH64: Missed L2I optimizations in C2 Summary: Use iRegIOrL2I for input operands whenever it makes sense. Reviewed-by: kvn changeset: 8151:2b0a471aea75 user: enevill date: Wed May 27 15:03:26 2015 +0100 files: src/cpu/aarch64/vm/aarch64_ad.m4 description: Add copyright to aarch64_ad.m4 changeset: 8150:5ecfe4a2327e user: aph date: Tue Mar 17 14:03:05 2015 +0000 files: src/cpu/aarch64/vm/macroAssembler_aarch64.hpp src/cpu/aarch64/vm/templateInterpreter_aarch64.cpp description: 8075045: AARCH64: Stack banging should use store rather than load Summary: Change stack bangs to use a store rather than a load Reviewed-by: dholmes changeset: 8149:fa858e3ae6f9 user: aph date: Fri Mar 13 12:44:28 2015 +0000 files: src/cpu/aarch64/vm/c1_LIRAssembler_aarch64.cpp description: 8074723: AARCH64: Stray pop in C1 LIR_Assembler::emit_profile_type Summary: Remove stray POP instruction Reviewed-by: dholmes changeset: 8148:20adeb715ada user: aph date: Tue Mar 03 17:56:33 2015 +0000 files: src/cpu/aarch64/vm/aarch64.ad description: 8074349: AARCH64: C2 generates poor code for some byte and character stores Summary: Use iRegIorL2I as src input for char and byte stores. Reviewed-by: kvn changeset: 8147:6d5b61ae5a7e user: aph date: Wed Jan 21 14:38:48 2015 -0800 files: src/cpu/aarch64/vm/sharedRuntime_aarch64.cpp src/cpu/aarch64/vm/templateInterpreter_aarch64.cpp src/share/vm/runtime/thread.hpp src/share/vm/runtime/thread.inline.hpp description: 8069593: Changes to JavaThread::_thread_state must use acquire and release Reviewed-by: kvn, dlong --- CUT HERE ---

4 7

RFR: AARCH64: The fix for hotspot's TestStable* tests
by Alexeev, Alexander 29 May '15

29 May '15

Hello This is porting of the fix for the series of tests from http://hg.openjdk.java.net/aarch64-port/jdk8/ to http://hg.openjdk.java.net/jdk9/dev The patch [1] corrects approach for method's compilation level determination with aarch64 specifics. Detailed description is provided as comments in the patch. Could somebody proceed with this? Review and sponsor if approved. Affected tests: compiler/stable/TestStableByte.java compiler/stable/TestStableBoolean.java compiler/stable/TestStableChar.java compiler/stable/TestStableFloat.java compiler/stable/TestStableObject.java compiler/stable/TestStableDouble.java compiler/stable/TestStableInt.java compiler/stable/TestStableLong.java compiler/stable/TestStableShort.java Before: Test results: passed: 817; failed: 35; error: 3 After: Test results: passed: 827; failed: 25; error: 3 [1] https://7013ccc350c4a4c1674f506214b6b6107b9d0df1.googledrive.com/host/0B5VQ…

1 0

JTREG, SPECjbb2013 and Hadoop/Terasort results for OpenJDK 8 on AArch64
by Edward Nevill 29 May '15

29 May '15

This is a summary of the JTREG test results =========================================== The build and test results are cycled every 10 days. For detailed information on the test output please refer to: http://openjdk.linaro.org/openjdk8-jtreg-nightly-tests/summary/2015/149/sum… ------------------------------------------------------------------------------- client-release/hotspot ------------------------------------------------------------------------------- Build 0: aarch64/2014/dec/05 pass: 584; fail: 53; error: 1 Build 1: aarch64/2014/dec/12 pass: 586; fail: 51; error: 1 Build 2: aarch64/2014/dec/16 pass: 585; fail: 52; error: 1 Build 3: aarch64/2015/jan/08 pass: 584; fail: 53; error: 1 Build 4: aarch64/2015/jan/09 pass: 584; fail: 53; error: 1 Build 5: aarch64/2015/feb/05 pass: 605; fail: 55; error: 2 Build 6: aarch64/2015/feb/13 pass: 605; fail: 55; error: 2 Build 7: aarch64/2015/mar/06 pass: 606; fail: 54; error: 2 Build 8: aarch64/2015/apr/17 pass: 605; fail: 55; error: 2 Build 9: aarch64/2015/may/29 pass: 605; fail: 55; error: 2 ------------------------------------------------------------------------------- client-release/jdk ------------------------------------------------------------------------------- Build 0: aarch64/2014/dec/05 pass: 5,323; fail: 217; error: 20 Build 1: aarch64/2014/dec/12 pass: 5,334; fail: 206; error: 20 Build 2: aarch64/2014/dec/16 pass: 5,332; fail: 205; error: 23 Build 3: aarch64/2015/jan/08 pass: 5,327; fail: 211; error: 22 Build 4: aarch64/2015/jan/09 pass: 5,330; fail: 212; error: 18 Build 5: aarch64/2015/feb/05 pass: 5,347; fail: 222; error: 19 Build 6: aarch64/2015/feb/13 pass: 5,357; fail: 215; error: 16 Build 7: aarch64/2015/mar/06 pass: 5,354; fail: 216; error: 18 Build 8: aarch64/2015/apr/17 pass: 5,359; fail: 208; error: 24 Build 9: aarch64/2015/may/29 pass: 5,350; fail: 217; error: 24 ------------------------------------------------------------------------------- client-release/langtools ------------------------------------------------------------------------------- Build 0: aarch64/2014/dec/05 pass: 3,037; error: 11 Build 1: aarch64/2014/dec/12 pass: 3,036; error: 12 Build 2: aarch64/2014/dec/16 pass: 3,039; error: 9 Build 3: aarch64/2015/jan/08 pass: 3,038; error: 10 Build 4: aarch64/2015/jan/09 pass: 3,037; error: 11 Build 5: aarch64/2015/feb/05 pass: 3,059; error: 14 Build 6: aarch64/2015/feb/13 pass: 3,064; error: 9 Build 7: aarch64/2015/mar/06 pass: 3,063; error: 10 Build 8: aarch64/2015/apr/17 pass: 3,063; error: 10 Build 9: aarch64/2015/may/29 pass: 3,062; error: 11 ------------------------------------------------------------------------------- server-release/hotspot ------------------------------------------------------------------------------- Build 0: aarch64/2014/dec/05 pass: 599; fail: 38; error: 1 Build 1: aarch64/2014/dec/12 pass: 599; fail: 38; error: 1 Build 2: aarch64/2014/dec/16 pass: 602; fail: 35; error: 1 Build 3: aarch64/2015/jan/08 pass: 599; fail: 38; error: 1 Build 4: aarch64/2015/jan/09 pass: 599; fail: 38; error: 1 Build 5: aarch64/2015/feb/05 pass: 624; fail: 32; error: 6 Build 6: aarch64/2015/feb/13 pass: 626; fail: 32; error: 4 Build 7: aarch64/2015/mar/06 pass: 626; fail: 32; error: 4 Build 8: aarch64/2015/apr/17 pass: 625; fail: 32; error: 5 Build 9: aarch64/2015/may/29 pass: 628; fail: 32; error: 2 ------------------------------------------------------------------------------- server-release/jdk ------------------------------------------------------------------------------- Build 0: aarch64/2014/dec/05 pass: 5,337; fail: 198; error: 25 Build 1: aarch64/2014/dec/12 pass: 5,318; fail: 218; error: 24 Build 2: aarch64/2014/dec/16 pass: 5,331; fail: 205; error: 24 Build 3: aarch64/2015/jan/08 pass: 5,330; fail: 202; error: 28 Build 4: aarch64/2015/jan/09 pass: 5,339; fail: 201; error: 20 Build 5: aarch64/2015/feb/05 pass: 5,349; fail: 209; error: 30 Build 6: aarch64/2015/feb/13 pass: 5,344; fail: 210; error: 34 Build 7: aarch64/2015/mar/06 pass: 5,348; fail: 207; error: 33 Build 8: aarch64/2015/apr/17 pass: 5,354; fail: 208; error: 29 Build 9: aarch64/2015/may/29 pass: 5,358; fail: 209; error: 24 ------------------------------------------------------------------------------- server-release/langtools ------------------------------------------------------------------------------- Build 0: aarch64/2014/dec/05 pass: 3,041; error: 7 Build 1: aarch64/2014/dec/12 pass: 3,040; error: 8 Build 2: aarch64/2014/dec/16 pass: 3,036; error: 12 Build 3: aarch64/2015/jan/08 pass: 3,038; fail: 1; error: 9 Build 4: aarch64/2015/jan/09 pass: 3,038; error: 10 Build 5: aarch64/2015/feb/05 pass: 3,062; error: 11 Build 6: aarch64/2015/feb/13 pass: 3,064; error: 9 Build 7: aarch64/2015/mar/06 pass: 3,065; error: 8 Build 8: aarch64/2015/apr/17 pass: 3,063; error: 10 Build 9: aarch64/2015/may/29 pass: 3,063; fail: 1; error: 9 Previous results can be found here: http://openjdk.linaro.org/openjdk8-jtreg-nightly-tests/index.html SPECjbb2013 composite regression test completed =============================================== This test measures the relative performance of the server compiler running the SPECjbb2013 composite tests and compares the performance against the baseline performance of the server compiler taken on 2014-04-01. In accordance with [1], the SPECjbb2013 tests are run on a system which is not production ready and does not meet all the requirements for publishing compliant results. The numbers below shall be treated as non-compliant (nc) and are for experimental purposes only. Relative performance: Server max-jOPS (nc): 1.21x Relative performance: Server critical-jOPS (nc): 1.69x Details of the test setup and historical results may be found here: http://openjdk.linaro.org/SPECjbb2013-1.00-results/ [1] http://www.spec.org/fairuse.html#Academic Regression test Hadoop-Terasort completed ========================================= This test measures the performance of the server and client compilers running Hadoop sorting a 1GB file using Terasort and compares the performance against the baseline performance of the Zero interpreter and against the baseline performance of the client and server compilers on 2014-04-01. Relative performance: Zero: 1.0, Client: 49.88, Server: 85.03 Client 49.88 / Client 2014-04-01 (43.00): 1.16x Server 85.03 / Server 2014-04-01 (71.00): 1.20x Details of the test setup and historical results may be found here: http://openjdk.linaro.org/hadoop-terasort-benchmark-results/ This is a summary of the jcstress test results ============================================== The build and test results are cycled every 10 days. 2014-12-05 pass rate: 11550/11550, results: http://openjdk.linaro.org/jcstress-nightly-runs/2014/339/results/ 2014-12-12 pass rate: 11550/11550, results: http://openjdk.linaro.org/jcstress-nightly-runs/2014/346/results/ 2014-12-16 pass rate: 11550/11550, results: http://openjdk.linaro.org/jcstress-nightly-runs/2014/350/results/ 2015-01-08 pass rate: 11550/11550, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/008/results/ 2015-01-09 pass rate: 11550/11550, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/009/results/ 2015-02-05 pass rate: 11551/11551, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/036/results/ 2015-02-13 pass rate: 11551/11551, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/044/results/ 2015-03-06 pass rate: 11551/11551, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/065/results/ 2015-04-17 pass rate: 11555/11555, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/107/results/ 2015-05-29 pass rate: 11556/11556, results: http://openjdk.linaro.org/jcstress-nightly-runs/2015/149/results/ For detailed information on the test output please refer to: http://openjdk.linaro.org/jcstress-nightly-runs/

1 0

hg: aarch64-port/jdk8/hotspot: 10 new changesets
by edward.nevill＠gmail.com 27 May '15

27 May '15

Changeset: 6d5b61ae5a7e Author: aph Date: 2015-01-21 14:38 -0800 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/6d5b61ae5a7e 8069593: Changes to JavaThread::_thread_state must use acquire and release Reviewed-by: kvn, dlong ! src/cpu/aarch64/vm/sharedRuntime_aarch64.cpp ! src/cpu/aarch64/vm/templateInterpreter_aarch64.cpp ! src/share/vm/runtime/thread.hpp ! src/share/vm/runtime/thread.inline.hpp Changeset: 20adeb715ada Author: aph Date: 2015-03-03 17:56 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/20adeb715ada 8074349: AARCH64: C2 generates poor code for some byte and character stores Summary: Use iRegIorL2I as src input for char and byte stores. Reviewed-by: kvn ! src/cpu/aarch64/vm/aarch64.ad Changeset: fa858e3ae6f9 Author: aph Date: 2015-03-13 12:44 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/fa858e3ae6f9 8074723: AARCH64: Stray pop in C1 LIR_Assembler::emit_profile_type Summary: Remove stray POP instruction Reviewed-by: dholmes ! src/cpu/aarch64/vm/c1_LIRAssembler_aarch64.cpp Changeset: 5ecfe4a2327e Author: aph Date: 2015-03-17 14:03 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/5ecfe4a2327e 8075045: AARCH64: Stack banging should use store rather than load Summary: Change stack bangs to use a store rather than a load Reviewed-by: dholmes ! src/cpu/aarch64/vm/macroAssembler_aarch64.hpp ! src/cpu/aarch64/vm/templateInterpreter_aarch64.cpp Changeset: 2b0a471aea75 Author: enevill Date: 2015-05-27 15:03 +0100 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/2b0a471aea75 Add copyright to aarch64_ad.m4 ! src/cpu/aarch64/vm/aarch64_ad.m4 Changeset: 5d3f35c13442 Author: aph Date: 2015-03-20 17:39 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/5d3f35c13442 8075443: AARCH64: Missed L2I optimizations in C2 Summary: Use iRegIOrL2I for input operands whenever it makes sense. Reviewed-by: kvn ! src/cpu/aarch64/vm/aarch64.ad ! src/cpu/aarch64/vm/aarch64_ad.m4 Changeset: 394a87600c41 Author: enevill Date: 2015-04-24 11:01 +0000 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/394a87600c41 8075930: AARCH64: Use FP Register in C2 Summary: modify to allow C2 to allocate FP (R29) as a general register Reviewed-by: aph, kvn, dlong ! src/cpu/aarch64/vm/aarch64.ad ! src/cpu/aarch64/vm/frame_aarch64.inline.hpp Changeset: e84a2db0758d Author: aph Date: 2015-04-14 17:19 +0100 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/e84a2db0758d 8076467: AARCH64: assertion fail with -XX:+UseG1GC Summary: Don't call encoding unless bool is true. Reviewed-by: kvn ! src/cpu/aarch64/vm/interp_masm_aarch64.hpp ! src/cpu/aarch64/vm/register_aarch64.hpp Changeset: adcffd0e1707 Author: enevill Date: 2015-05-27 15:28 +0100 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/adcffd0e1707 8079203: AARCH64: Need to cater for different partner implementations Summary: Parse /proc/cpuinfo to derive implementation specific info ! src/cpu/aarch64/vm/aarch64.ad ! src/cpu/aarch64/vm/vm_version_aarch64.cpp ! src/cpu/aarch64/vm/vm_version_aarch64.hpp Changeset: d38b6415fcd8 Author: enevill Date: 2015-05-27 15:40 +0100 URL: http://hg.openjdk.java.net/aarch64-port/jdk8/hotspot/rev/d38b6415fcd8 8080586: aarch64: hotspot test compiler/codegen/7184394/TestAESMain.java fails Summary: Return correct length in generate_cipherBlockChaining_encryptAESCrypt ! src/cpu/aarch64/vm/stubGenerator_aarch64.cpp

1 0

Where GrinderBench benchmark can be downloaded
by Alexeev, Alexander 27 May '15

27 May '15

Hello I saw that GrinderBench benchmark is used from time to time for JVM performance evaluation on aarch64. EEMBC site [1] says that Grinderbench is no longer supported and no download link is available. Where can I get GrinderBench binaries for execution? Thanks, Alexander [1] http://www.eembc.org/grinderbench/

1 0

RFR: AARCH64: RewriteFrequentPairs optimization
by Alexeev, Alexander 27 May '15

27 May '15

Hello Could somebody review the patch [1] and sponsor if approved? Patch implements RewriteFrequentPairs optimization for aarch64. It was tested with hotspot's jtreg test suite and DaCapo benchmark avrora executed in interpretation mode. Hotspot tests: Before: Test results: passed: 801; failed: 47; error: 1 After Test results: passed: 801; failed: 47; error: 1 Dacapo: $java -Xint -XX:+RewriteFrequentPairs -jar dacapo-9.12-bach.jar avrora ===== DaCapo 9.12 avrora starting ===== ===== DaCapo 9.12 avrora PASSED in 45085 msec ===== [1] https://googledrive.com/host/0B5VQvD5uJjDQfnF4eVJYT1pPS2UxQ0N1YUJNQ0VBZy1SM…

2 1

[9] RFR (XS) 8051712 regression Test7107135 crashes
by Chris Plummer 26 May '15

26 May '15

Hello, Please review the fix for 8051712: http://cr.openjdk.java.net/~cjplummer/8051712/webrev.00 https://bugs.openjdk.java.net/browse/JDK-8051712 This fix was already applied to aarch64-port/jdk8, but was missed when aarch64-port was integrated into jdk9. This commit is to the jdk9/dev mainline, not aarch64-port/jdk9. The issue is that ElfFile::specifies_noexecstack() assumes that if no PT_GNU_STACK record is found, then the platform (loader) defaults to execstack. However, for AARCH64 is defaults to noexecstack. More info can be found in the following discussion. http://mail.openjdk.java.net/pipermail/aarch64-port-dev/2014-February/00087… Also see 7107135 for info on the execstack/noexecstack issue, and why ElfFile::specifies_noexecstack() exists in the first place: https://bugs.openjdk.java.net/browse/JDK-7107135 Tested by using JPRT to run the test case on all supported platforms. Also ran all default "-testset hotspot" tests on all platforms. thanks, Chris

4 4