[11u] RFR (M) 8251525: AARCH64: Faster Math.signum(fp)

[Hohensee, Paul]

Lgtm to both.

Thanks,
Paul

-----Original Message-----
From: jdk-updates-dev <jdk-updates-dev-retn at openjdk.java.net> on behalf of Dmitry Chuyko <dmitry.chuyko at bell-sw.com>
Date: Thursday, April 15, 2021 at 10:15 AM
To: "jdk-updates-dev at openjdk.java.net" <jdk-updates-dev at openjdk.java.net>
Subject: [11u] RFR (M) 8251525: AARCH64: Faster Math.signum(fp)

Hello,

Original RFE: https://bugs.openjdk.java.net/browse/JDK-8251525
Original post-fix: https://bugs.openjdk.java.net/browse/JDK-8252779

The patch required minor adjustments:

src/hotspot/share/opto/intrinsicnode.hpp
Copyright change.

src/hotspot/share/opto/library_call.cpp
Case additions with missing context are reproduced in
LibraryCallKit::inline_double_math() (JDK-8231649 is missing).

src/hotspot/share/opto/matcher.cpp
Additions with missing context are reproduced in Matcher::find_shared()
(Matcher::find_shared_post_visit(), JDK-8213746 is missing)

Initial post-fix for CheckGraalIntrinsics (HotspotTest) Graal test was
changed to mark new intrinsics as toBeInvestigated in JDK 11+.

11u webrev signum:
http://cr.openjdk.java.net/~dchuyko/8251525/webrev.11u.00/
11u webrev post-fix:
http://cr.openjdk.java.net/~dchuyko/8252779/webrev.11u.00/

Testing: compiler/intrinsics/math/TestSignumIntrinsic.java, tier1,
tier2; compiler/graalunit with Graal on - CheckGraalIntrinsics fails
because of ECB intrinsics (JDK-8229848).

--
Thanks,
-Dmitry