May 2021 - core-libs-dev

JDK Dynalink only works with unconditionally exported packages
by Thiago Henrique Hupner 10 May '21

10 May '21

Hi all. I've been testing the JDK Dynalink recently and I think I've found a bug. The class jdk.dynalink.beans.CheckRestrictedPackage checks if a package is restricted. So, I have a class that has some static methods. But to Dynalink find the class, the class needs to be public. OK, I've changed my class to be public. However, it needs that the package of the class is exported in the module too. I think this is a little too restrictive. I don't want to expose to my users some internal implementation. The main problem is that the method CheckRestrictedPackage::isRestrictedClass does the following check: // ... final String pkgName = name.substring(0, i); final Module module = clazz.getModule(); if (module != null && !module.isExported(pkgName)) { return true; } // ... I have a feeling that this check should be changed to something like // ... final String pkgName = name.substring(0, i); final Module module = clazz.getModule(); if (module != null && !module.isOpen(pkgName, CheckRestrictedPackage.class.getModule())) { return true; } // ... In this way, my code can only "opens" a single package to the jdk.dynalink module, not having to unconditionally export a package. >From what I could understand, in Nashorn, the generated classes were defined in the unnamed module, that always exports all the packages, so the access always worked. The code can be found here: https://github.com/openjdk/jdk/blob/master/src/jdk.dynalink/share/classes/j… Best regards. Thiago

1 0

Package-private members in java.util.ArrayList
by Zheka Kozlov 10 May '21

10 May '21

Hi! JEP 181 (Nestmates) was implemented in Java 11, however, there are still many library classes with members that were declared package-private to avoid synthetic bridge methods. For example, ArrayList and its friends have the following declarations: • transient Object[] elementData; // non-private to simplify nested class access • private class Itr implements Iterator<E> { int cursor; // index of next element to return int lastRet = -1; // index of last element returned; -1 if no such int expectedModCount = modCount; // prevent creating a synthetic constructor Itr() {} ... final void checkForComodification() { ... } } • private class ListItr extends Itr implements ListIterator<E> { ListItr(int index) { super(); cursor = index; } ... } • final class ArrayListSpliterator implements Spliterator<E> { ArrayListSpliterator(int origin, int fence, int expectedModCount) { this.index = origin; this.fence = fence; this.expectedModCount = expectedModCount; } ... } As far as I understand, this is no longer relevant and we can freely make all these members private without undermining performance. This will make the code more readable and will stop confusing readers. Can we do it? Zheka.

1 0

RFR: 8266610: Method RandomAccessFile#length() returns 0 for block devices on linux.
by Vyom Tewari 10 May '21

10 May '21

RandomAccessFile.length() method for block device return always 0 ------------- Commit messages: - JDK-8266610: Method RandomAccessFile#length() returns 0 for block devices on linux. Changes: https://git.openjdk.java.net/jdk/pull/3914/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk&pr=3914&range=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8266610 Stats: 13 lines in 1 file changed: 10 ins; 2 del; 1 mod Patch: https://git.openjdk.java.net/jdk/pull/3914.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/3914/head:pull/3914 PR: https://git.openjdk.java.net/jdk/pull/3914

5 10

Improvement of Dual-Pivot Quicksort (Radix sort)
by Vladimir Yaroslavskiy 08 May '21

08 May '21

Hi, I’d like to share new improvement of Dual-Pivot Quicksort: new algorithm of sorting, Radix sort, is suggested. Also performance was improved in case of corner merging of runs. The changes are: Sorting: - adopt radix sort for sequential and parallel sorts on int/long/float/double arrays (almost random and length > 6K) - fix tryMergeRuns() to better handle case when the last run is a single element - minor javadoc and comment changes Testing: - add new data inputs in tests for sorting - add min/max/infinity values to float/double testing - add tests for radix sort See related JDB and PR https://bugs.openjdk.java.net/browse/JDK-8266431 https://github.com/openjdk/jdk/pull/3938 Laurent Bourges has tested new version of Dual-Pivot Quicksort, it shows better performance. --- STATS --- Comparing sorters SYSTEM vs NEW_DPQS_REF winners : 75 / 197 avg (%): 125.5317 stats (%): [360: µ=146.05562 σ=106.212036 rms=252.26765 min=29.021154 max=618.46735] Stats per keys: IDENT_____:SPIRAL stats (%): [12: µ=146.85367 σ=67.929474 rms=214.78313 min=81.47126 max=301.9469] IDENT_____:STAGGER stats (%): [12: µ=124.82189 σ=85.498825 rms=210.32072 min=81.963005 max=395.48715] IDENT_____:SAWTOTH stats (%): [12: µ=121.015236 σ=48.471672 rms=169.48691 min=92.21825 max=227.12547] IDENT_____:_RANDOM stats (%): [12: µ=235.53735 σ=204.9386 rms=440.47595 min=84.627625 max=613.981] IDENT_____:PLATEAU stats (%): [12: µ=99.05677 σ=1.7938316 rms=100.85061 min=95.937035 max=100.65806] IDENT_____:SHUFFLE stats (%): [12: µ=91.41511 σ=19.296474 rms=110.71158 min=57.53845 max=117.839775] REVERSE___:SPIRAL stats (%): [12: µ=214.24744 σ=86.17577 rms=300.42322 min=90.77709 max=355.51056] REVERSE___:STAGGER stats (%): [12: µ=127.6578 σ=87.32062 rms=214.97841 min=94.24164 max=404.43817] REVERSE___:SAWTOTH stats (%): [12: µ=121.37123 σ=39.691704 rms=161.06294 min=95.34109 max=209.51929] REVERSE___:_RANDOM stats (%): [12: µ=251.44595 σ=221.17654 rms=472.6225 min=86.738144 max=618.46735] REVERSE___:PLATEAU stats (%): [12: µ=104.74188 σ=6.7017303 rms=111.44361 min=97.345634 max=120.3795] REVERSE___:SHUFFLE stats (%): [12: µ=100.18308 σ=19.42561 rms=119.608696 min=69.78542 max=135.48901] DITHER____:SPIRAL stats (%): [12: µ=183.42812 σ=64.67034 rms=248.09845 min=99.713776 max=268.3382] DITHER____:STAGGER stats (%): [12: µ=174.41586 σ=123.18825 rms=297.60413 min=99.123314 max=459.56564] DITHER____:SAWTOTH stats (%): [12: µ=187.5975 σ=110.62237 rms=298.21988 min=97.2715 max=367.14056] DITHER____:_RANDOM stats (%): [12: µ=185.65709 σ=129.0605 rms=314.7176 min=70.168724 max=434.8079] DITHER____:PLATEAU stats (%): [12: µ=100.72014 σ=3.9355173 rms=104.655655 min=94.559074 max=107.78911] DITHER____:SHUFFLE stats (%): [12: µ=86.45017 σ=42.46921 rms=128.91939 min=29.021154 max=171.31981] REVERSE_BA:SPIRAL stats (%): [12: µ=177.85292 σ=85.1786 rms=263.03152 min=94.477264 max=315.87576] REVERSE_BA:STAGGER stats (%): [12: µ=132.4794 σ=103.280464 rms=235.75987 min=97.76542 max=460.07227] REVERSE_BA:SAWTOTH stats (%): [12: µ=110.42876 σ=23.894398 rms=134.32315 min=98.04496 max=182.87247] REVERSE_BA:_RANDOM stats (%): [12: µ=223.70018 σ=185.21355 rms=408.91373 min=86.62554 max=618.0156] REVERSE_BA:PLATEAU stats (%): [12: µ=99.096275 σ=1.8192571 rms=100.91553 min=95.90717 max=100.805466] REVERSE_BA:SHUFFLE stats (%): [12: µ=101.3669 σ=22.24021 rms=123.60711 min=65.73001 max=144.21266] REVERSE_FR:SPIRAL stats (%): [12: µ=190.94035 σ=74.251236 rms=265.1916 min=97.7578 max=358.39563] REVERSE_FR:STAGGER stats (%): [12: µ=134.06009 σ=104.719696 rms=238.77979 min=98.797386 max=466.2626] REVERSE_FR:SAWTOTH stats (%): [12: µ=110.6682 σ=27.3231 rms=137.9913 min=94.27639 max=195.0114] REVERSE_FR:_RANDOM stats (%): [12: µ=248.0682 σ=201.4303 rms=449.4985 min=97.789764 max=614.1303] REVERSE_FR:PLATEAU stats (%): [12: µ=99.416336 σ=7.2677627 rms=106.6841 min=87.80682 max=115.78382] REVERSE_FR:SHUFFLE stats (%): [12: µ=96.97448 σ=18.399363 rms=115.37385 min=64.20668 max=130.8938] https://github.com/bourgesl/nearly-optimal-mergesort-code/tree/master/sort-… Radix Sort benchmark https://github.com/bourgesl/radix-sort-benchmark/blob/main/results/2021-ref… Thank you, Vladimir

1 0

RFR: 8266646: Add more diagnostic to java/lang/System/LoggerFinder/modules
by Daniel Fuchs 07 May '21

07 May '21

Hi, please find here a trivial test change that adds some diagnostic (time stamps) to the LoggerFinder/modules subprocess test logs. ------------- Commit messages: - 8266646: Add more diagnostic to java/lang/System/LoggerFinder/modules Changes: https://git.openjdk.java.net/jdk/pull/3904/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk&pr=3904&range=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8266646 Stats: 38 lines in 4 files changed: 27 ins; 2 del; 9 mod Patch: https://git.openjdk.java.net/jdk/pull/3904.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/3904/head:pull/3904 PR: https://git.openjdk.java.net/jdk/pull/3904

5 11

Re: RFR: 8265768 [aarch64] Use glibc libm impl for dlog, dlog10, dexp iff 2.29 or greater on AArch64.
by Carlos O'Donell 07 May '21

07 May '21

On Thu, 15 Apr 2021 08:33:47 GMT, gregcawthorne <github.com+73799211+gregcawthorne(a)openjdk.org> wrote: > Glibc 2.29 onwards provides optimised versions of log,log10,exp. > These functions have an accuracy of 0.9ulp or better in glibc > 2.29. > > Therefore this patch adds code to parse, store and check > the runtime glibcs version in os_linux.cpp/hpp. > This is then used to select the glibcs implementation of > log, log10, exp at runtime for c1 and c2, iff we have > glibc 2.29 or greater. > > This will ensure OpenJDK can benefit from future improvements > to glibc. > > Glibc adheres to the ieee754 standard, unless stated otherwise > in its spec. > > As there are no stated exceptions in the current glibc spec > for dlog, dlog10 and dexp, we can assume they currently follow > ieee754 (which testing confirms). As such, future version of > glibc are unlikely to lose this compliance with ieee754 in > future. > > W.r.t performance this patch sees ~15-30% performance improvements for > log and log10, with ~50-80% performance improvements for exp for the > common input ranged (which output real numbers). However for the NaN > and inf output ranges we see a slow down of up to a factor of 2 for > some functions and architectures. > > Due to this being the uncommon case we assert that this is a > worthwhile tradeoff. Where does the requirement for monotonicity come from? ------------- PR: https://git.openjdk.java.net/jdk/pull/3510

3 2

RFR: 8265465: jcmd VM.cds should keep already dumped archive when exceptions happens
by Yumin Qi 07 May '21

07 May '21

Hi, Please review When using jcmd to dump shared archive, if the archive name exists, it will be deleted first. If exception happens during dumping process, there is no new archive created. This PR changes to first dump the archive with a temporary file name. With successful dump, the temporary file will be rename to its given name. This way the old archive will not be touched on exception. The newly added test case skipped testing on Windows since File operation result is not same as on Linux. Tests: tier1,tier2,tier3,tier4 Thanks Yumin ------------- Commit messages: - 8265465: jcmd VM.cds should keep already dumped archive when exception happens Changes: https://git.openjdk.java.net/jdk/pull/3886/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk&pr=3886&range=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8265465 Stats: 188 lines in 3 files changed: 158 ins; 5 del; 25 mod Patch: https://git.openjdk.java.net/jdk/pull/3886.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/3886/head:pull/3886 PR: https://git.openjdk.java.net/jdk/pull/3886

3 13

New candidate JEP: 415: Context-Specific Deserialization Filters
by mark.reinhold＠oracle.com 06 May '21

06 May '21

https://openjdk.java.net/jeps/415 Summary: Allow applications to configure context-specific and dynamically-selected deserialization filters via a JVM-wide filter factory that is invoked to select a filter for each individual deserialization operation. - Mark

1 0

RFR: 8266225: jarsigner is using incorrect security property to show weakness of certs
by Hai-May Chao 06 May '21

06 May '21

Please review the change to jarsigner so it uses certpath security property in order to properly display the weakness of the certificate algorithms. ------------- Commit messages: - 8266225:jarsigner is using incorrect security property to show weakness of certs Changes: https://git.openjdk.java.net/jdk/pull/3905/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk&pr=3905&range=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8266225 Stats: 39 lines in 2 files changed: 31 ins; 0 del; 8 mod Patch: https://git.openjdk.java.net/jdk/pull/3905.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/3905/head:pull/3905 PR: https://git.openjdk.java.net/jdk/pull/3905

1 0

RFR: 8266460: java.io tests fail on null stream with upgraded jtreg/TestNG
by Lance Andersen 06 May '21

06 May '21

Hi all, This fix addresses a change in TestNG behavior for the Before/AfterGroups annotations that was introduced via https://github.com/cbeust/testng/pull/2167. The tests have been verified against the latest TestNG release and continue to run with the current TestNG release used by jtreg ------------- Commit messages: - remove trailing space - Updates for TestNG 7.4 Changes: https://git.openjdk.java.net/jdk/pull/3866/files Webrev: https://webrevs.openjdk.java.net/?repo=jdk&pr=3866&range=00 Issue: https://bugs.openjdk.java.net/browse/JDK-8266460 Stats: 122 lines in 4 files changed: 12 ins; 24 del; 86 mod Patch: https://git.openjdk.java.net/jdk/pull/3866.diff Fetch: git fetch https://git.openjdk.java.net/jdk pull/3866/head:pull/3866 PR: https://git.openjdk.java.net/jdk/pull/3866

2 7