Request for approval: JDK-8073108 "Use x86 and SPARC CPU instructions for GHASH acceleration"

Yuri Gaevsky ygaevsky at azul.com
Fri Sep 27 14:50:38 UTC 2019


Hello,

Please approve backport of JDK-8073108 "Use x86 and SPARC CPU instructions for GHASH acceleration" to jdk8u.

Bug:
  https://bugs.openjdk.java.net/browse/JDK-8073108

JDK9 review thread:
  http://mail.openjdk.java.net/pipermail/hotspot-dev/2015-February/017178.html

Changesets:
  http://hg.openjdk.java.net/jdk9/jdk9/hotspot/rev/ce0c612ea443
  http://hg.openjdk.java.net/jdk9/jdk9/jdk/rev/1e10709d34af

Webrevs:
  http://cr.openjdk.java.net/~ygaevsky/8073108.8u/webrev.hotspot/
  http://cr.openjdk.java.net/~ygaevsky/8073108.8u/webrev.jdk/
The original changes for jdk9 do not apply cleanly to jdk8u codebase
due to slightly different file layouts and absence of aarch64 platform.

Testing:
  windows-i586: ( known failures due to https://bugs.openjdk.java.net/browse/JDK-8130341 (work in progress))
  windows-x64:
  linux-x64:
    JTREG tests: 
      (jdk8u) hotspot/test/compiler/7184394/
      (jdk8u) jdk/test/com/sun/crypto/provider/Cipher/AES/
    (also checked the output from TestAESMain for performance improvements w/ -XX:+UseGHASHIntrinsics)
    
  build for PowerPC:
    checked output for -XX:+UseGHASHIntrinsics ("GHASH intrinsics are not available on this CPU")

  build for Solaris-sparcv9:
    no JTREG tests were run because the available sparc machine doesn't support VIS3 instruction set,
    so just checked output for -XX:+UseGHASHIntrinsics ("GHASH intrinsics require VIS3 insructions
    support. Intriniscs will be disabled").
    It would be great if someone could check the real intrinsic code on the appropriate
    solaris-sparc (SPARC T4?) hardware.

Thank you,
Yuri Gaevsky
Azul Systems, Inc.



More information about the jdk8u-dev mailing list