September 2019 - aarch64-port-dev

RFR: 8218748: AARCH64: String::compareTo intrinsic documentation and maintenance improvement
by Dmitrij Pochepko 18 Nov '19

18 Nov '19

Hi all. Please review an enhancement change "8218748: AARCH64: String::compareTo intrinsic documentation and maintenance improvement". RFE: https://bugs.openjdk.java.net/browse/JDK-8218748 webrev: http://cr.openjdk.java.net/~dpochepk/8218748/webrev/ Changes: - Added documentation for compareTo implementation. - Removed duplicated code and using existing code instead. -- compare_string_16_bytes_same is used for tail handling in generate_compare_long_string_same_encoding . -- DIFF block code is reused for DIFF2 block in generate_compare_long_string_same_encoding. - Registers and labels are partially renamed to better reflect code actions. This patch is dependent on JDK-8218966 and should be applied on top of it. Testing: - jck - hotspot jtreg compiler/*, runtime/* and gc/* tests. Including AARCH64-implementation-specific compareTo tests introduced in JDK-8218966 fix - jdk jtreg tier1-3 tests No regressions were found. I'd like to thank Pengfei Li (Pengfei.Li(a)arm.com) for pre-review and additional testing. Thanks, Dmitrij

5 10

RFR (trivial) : fix aarch64-8u type profile bug
by Yangfei (Felix) 07 Nov '19

07 Nov '19

Hi, When I play with a simple test case (InstanceOfExample.java), I witnessed strange C2 JIT code generated by aarch64-8u: 000 B1: # B9 B2 <- BLOCK HEAD IS JUNK Freq: 1 000 # stack bang size=48 sub sp, sp, #48 stp rfp, lr, [sp, #32] 014 ldr R10, [R1, #8] # class 018 NullCheck R1 018 018 B2: # B8 B3 <- B1 Freq: 0.999999 018 + mov R4, precise klass MotorCycle: 0x0000ffff7c069030:Constant:exact * # ptr < ============== 024 + cmp R10, R4 // ptr < ============== 028 bne B8 # unsigned P=0.000000 C=-1.000000 < ============== 028 02c B3: # B6 B4 <- B2 Freq: 0.999999 02c + ldr R10, [R4, #32] # class 030 + mov R0, precise klass DriveCar: 0x0000ffff7c06a750:Constant:exact * # ptr 03c + cmp R10, R0 // ptr 040 bne B6 # unsigned P=0.170000 C=-1.000000 …… 0b0 B8: # N1 <- B2 Freq: 1e-35 0b0 + spill R1 -> [sp, #0] # spill size = 64 0b4 + mov R1, #-34 # int 0b8 call,static 0x0000ffff8979efc0 // ==> wrapper for: uncommon_trap(reason='class_check' action='maybe_recompile') # InstanceOfExample::instanceOfDriverCar @ bci:1 L[0]=_ STK[0]=sp + #0 # OopMap{[0]=Oop off=188} 0bc ShouldNotReachHere Here, C2 inserts an uncommon trap in B2 if the incoming parameter ‘o’ of instanceOfDriverCar method is not an instance of MotorCycle. As we send three different types of objects to instanceOfDriverCar method, this JIT code is not expected. The uncommon trap happens after Thread.sleep in main when instanceOfDriverCar got a Car object. GDB shows that profile_has_unique_klass gives a wrong result: profile.morphism() equals 1, but it should be 2 here, and this misleads the C2 compiler. I found this bug is fixed in: http://hg.openjdk.java.net/jdk9/jdk9/hotspot/rev/22fd02fad88b Fix for aarch64-8u is trivial: diff -r a1986eb5d5fd src/cpu/aarch64/vm/interp_masm_aarch64.cpp --- a/src/cpu/aarch64/vm/interp_masm_aarch64.cpp Mon Sep 09 18:54:16 2019 +0200 +++ b/src/cpu/aarch64/vm/interp_masm_aarch64.cpp Fri Sep 20 11:09:10 2019 +0800 @@ -1163,12 +1163,12 @@ b(done); bind(found_null); } else { - cbz(reg2, done); + cbnz(reg2, done); } break; } // Since null is rare, make it be the branch-taken case. - cbz(reg2,found_null); + cbz(reg2, found_null); // Put all the "Case 3" tests here. record_klass_in_profile_helper(receiver, mdp, reg2, start_row + 1, done, is_virtual_call); Is it OK to go? Thanks. Command line: java -XX:+UseSerialGC -Xms128M -Xmx128M -XX:-UseCompressedOops -XX:-UseCompressedClassPointers -XX:CICompilerCount=2 -XX:+PrintCompilation -XX:-TieredCompilation -XX:LoopMaxUnroll=0 -XX:CompileOnly=InstanceOfExample::instanceOfDriverCar -XX:-Inline -XX:+UnlockDiagnosticVMOptions -XX:+PrintOptoAssembly InstanceOfExample InstanceOfExample.java: public class InstanceOfExample { public static int count = 0; public static void main(String[] args) throws Exception { Vehicle vehicle = new Vehicle(); Car car = new Car(); MotorCycle moto = new MotorCycle(); Object[] tools = new Object[]{vehicle, car, moto}; for (int i = 0; i < 20000; i++) { if (instanceOfDriverCar(tools[i % 3])) { count++; } } Thread.sleep(2000); if (instanceOfDriverCar(car)) { count++; } } public static boolean instanceOfDriverCar(Object o) { if (o instanceof DriveCar) { return true; } return false; } } class Vehicle { } class Car extends Vehicle implements DriveCar { } class MotorCycle extends Vehicle { } interface DriveCar { }

2 9

[8u] 8231366: Shenandoah: Shenandoah String Dedup thread is not properly initialized
by Zhengyu Gu 04 Nov '19

04 Nov '19

This bug seems to exist since day one of 8u backport. The ConcurrentGCThread API is different in 8u and we leave ShenandoahDedupThread not properly initialized before it enters work loop. In Shenandoah String Deduplication tests, the bug results assertion failure that shows Thread::current() == NULL. The bug only manifests on Windows, is due to discrepancy of java_start() implementation on different OSs. e.g. it sets *thread* on Linux. Bug: https://bugs.openjdk.java.net/browse/JDK-8231366 Webrev: http://cr.openjdk.java.net/~zgu/JDK-8231366/webrev.00/ Test: hotspot_gc_shenandoah on Windows and Linux. Thanks, -Zhengyu

4 6

Re: [aarch64-port-dev ] RFR: [8u] 8209835: Aarch64: elide barriers on all volatile operations
by Yangfei (Felix) 31 Oct '19

31 Oct '19

On 8/28/19 10:18 AM, Yangfei (Felix) wrote: >> Well, I think I won't be able to provide such a test case after some >> throughts. Now I agree that this will not affect basic functionality. >> As we have handled unsafe.compareAndSwap, I don't see a reason to not >> handle unsafe.getAndSet: the risk should be low and we get consistent >> and improved C2 JIT code. What do you think? > I think that's a powerful argument for the development code because it's an improvement; conversely I don't think it's an argument for a backport. > Having said that, I can sort-of imagine a jcstress failure if we had a test that was doing compareAndSwap in one thread and getAndSet in another. That would be a seriously compelling reason for a backport! Can you elaborate more please? especially the possible code sequence. Currently, we have C2 JIT for getAndSet like this: 33 ;; membar_release 34 0x0000ffffa416ae74: dmb ish 35 ;; 0x40000000000 36 0x0000ffffa416ae78: orr x11, xzr, #0x40000000000 37 0x0000ffffa416ae7c: add x10, x1, #0x10 38 0x0000ffffa416ae80: prfm pstl1strm, [x10] 39 0x0000ffffa416ae84: ldxr x9, [x10] 40 0x0000ffffa416ae88: stxr w8, x11, [x10] 41 0x0000ffffa416ae8c: cbnz w8, 0x0000ffffa416ae84 42 0x0000ffffa416ae90: mov x10, x9 43 ;; membar_acquire 44 0x0000ffffa416ae94: dmb ishld ;*invokevirtual getAndSetLong I think the two dmb instructions at the head and tail should ensure the consistency. Isn't it? > It would be interesting to see such a jcstress test. Even if the test does not fail, we could look at the generated code to see if it might fail according to the Arm memory model.

3 11

hg: aarch64-port/jdk8u-shenandoah/nashorn: 3 new changesets
by gnu.andrew＠redhat.com 30 Sep '19

30 Sep '19

Changeset: fb70ff525e04 Author: andrew Date: 2019-09-23 06:13 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/nashorn/rev/fb70f… Added tag jdk8u232-b07 for changeset 0f61e27241b5 ! .hgtags Changeset: a88c90879254 Author: andrew Date: 2019-09-24 03:29 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/nashorn/rev/a88c9… Merge jdk8u232-b07 ! .hgtags Changeset: 2cf6a7658b5c Author: andrew Date: 2019-09-24 03:30 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/nashorn/rev/2cf6a… Added tag aarch64-shenandoah-jdk8u232-b07 for changeset a88c90879254 ! .hgtags

1 0

hg: aarch64-port/jdk8u-shenandoah/jdk: 3 new changesets
by gnu.andrew＠redhat.com 30 Sep '19

30 Sep '19

Changeset: ba6a242c5223 Author: andrew Date: 2019-09-23 06:13 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jdk/rev/ba6a242c5… Added tag jdk8u232-b07 for changeset 3560e0ebe876 ! .hgtags Changeset: 63d584bd574f Author: andrew Date: 2019-09-24 03:29 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jdk/rev/63d584bd5… Merge jdk8u232-b07 ! .hgtags Changeset: 282353338f2d Author: andrew Date: 2019-09-24 03:30 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jdk/rev/282353338… Added tag aarch64-shenandoah-jdk8u232-b07 for changeset 63d584bd574f ! .hgtags

1 0

hg: aarch64-port/jdk8u-shenandoah/hotspot: 4 new changesets
by gnu.andrew＠redhat.com 30 Sep '19

30 Sep '19

Changeset: c751303497d5 Author: dsamersoff Date: 2014-06-13 05:10 -0700 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/hotspot/rev/c7513… 8038392: Generating prelink cache breaks JAVA 'jinfo' utility normal behaviour Summary: Better parsing of /proc/pid/maps in sa Reviewed-by: sspitsyn, sla ! agent/src/os/linux/ps_proc.c Changeset: 1aa54ae51b56 Author: andrew Date: 2019-09-23 06:13 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/hotspot/rev/1aa54… Added tag jdk8u232-b07 for changeset c751303497d5 ! .hgtags Changeset: 3a72593f3e4b Author: andrew Date: 2019-09-24 03:29 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/hotspot/rev/3a725… Merge jdk8u232-b07 ! .hgtags ! agent/src/os/linux/ps_proc.c Changeset: b8c0849c1d94 Author: andrew Date: 2019-09-24 03:30 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/hotspot/rev/b8c08… Added tag aarch64-shenandoah-jdk8u232-b07 for changeset 3a72593f3e4b ! .hgtags

1 0

hg: aarch64-port/jdk8u-shenandoah/langtools: 3 new changesets
by gnu.andrew＠redhat.com 30 Sep '19

30 Sep '19

Changeset: 4522596811bb Author: andrew Date: 2019-09-23 06:13 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/langtools/rev/452… Added tag jdk8u232-b07 for changeset 3b3a43588afb ! .hgtags Changeset: e113b2820cc8 Author: andrew Date: 2019-09-24 03:29 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/langtools/rev/e11… Merge jdk8u232-b07 ! .hgtags Changeset: 5ce4c7045a9f Author: andrew Date: 2019-09-24 03:30 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/langtools/rev/5ce… Added tag aarch64-shenandoah-jdk8u232-b07 for changeset e113b2820cc8 ! .hgtags

1 0

hg: aarch64-port/jdk8u-shenandoah/jaxws: 3 new changesets
by gnu.andrew＠redhat.com 30 Sep '19

30 Sep '19

Changeset: bd2b62946fe7 Author: andrew Date: 2019-09-23 06:13 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jaxws/rev/bd2b629… Added tag jdk8u232-b07 for changeset fb3b24b266a2 ! .hgtags Changeset: 9bce6fb57b44 Author: andrew Date: 2019-09-24 03:28 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jaxws/rev/9bce6fb… Merge jdk8u232-b07 ! .hgtags Changeset: 8116b048aa1c Author: andrew Date: 2019-09-24 03:30 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jaxws/rev/8116b04… Added tag aarch64-shenandoah-jdk8u232-b07 for changeset 9bce6fb57b44 ! .hgtags

1 0

hg: aarch64-port/jdk8u-shenandoah/jaxp: 3 new changesets
by gnu.andrew＠redhat.com 30 Sep '19

30 Sep '19

Changeset: 09bc39a2d2d2 Author: andrew Date: 2019-09-23 06:13 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jaxp/rev/09bc39a2… Added tag jdk8u232-b07 for changeset 0a5bd84fd43f ! .hgtags Changeset: 4d61ecb283d2 Author: andrew Date: 2019-09-24 03:28 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jaxp/rev/4d61ecb2… Merge jdk8u232-b07 ! .hgtags Changeset: ca269f9523f5 Author: andrew Date: 2019-09-24 03:30 +0100 URL: https://hg.openjdk.java.net/aarch64-port/jdk8u-shenandoah/jaxp/rev/ca269f95… Added tag aarch64-shenandoah-jdk8u232-b07 for changeset 4d61ecb283d2 ! .hgtags

1 0