January 2025 - hotspot-jfr-dev

[jdk24] RFR: 8343510: JFR: Remove AccessControlContext from FlightRecorder::addListener specification
by Erik Gahlin 14 Jan '25

14 Jan '25

8343510: JFR: Remove AccessControlContext from FlightRecorder::addListener specification ------------- Commit messages: - Backport 1bf2f5c8a92b30eabb530737158f57c63a81fef6 Changes: https://git.openjdk.org/jdk/pull/23041/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=23041&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8343510 Stats: 2 lines in 1 file changed: 0 ins; 1 del; 1 mod Patch: https://git.openjdk.org/jdk/pull/23041.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/23041/head:pull/23041 PR: https://git.openjdk.org/jdk/pull/23041

2 2

RFR: 8347724: Replace SIZE_FORMAT in jfr directory
by Coleen Phillimore 14 Jan '25

14 Jan '25

Please review this change to replace SIZE_FORMAT with %zu in the jfr directory. Most were done with a script with no hand-editing. Tested just now with tier1-4 on x86 and aarch64. ------------- Commit messages: - Round 1 with the script. Changes: https://git.openjdk.org/jdk/pull/23109/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=23109&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8347724 Stats: 28 lines in 9 files changed: 0 ins; 0 del; 28 mod Patch: https://git.openjdk.org/jdk/pull/23109.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/23109/head:pull/23109 PR: https://git.openjdk.org/jdk/pull/23109

2 5

RFR: 8347496: Test jdk/jfr/jvm/TestModularImage.java fails after JDK-8347124: No javac
by Severin Gehwolf 14 Jan '25

14 Jan '25

Please review this test-only fix. Apparently there are some test configurations out there that run `jdk/jfr/jvm/TestModularImage.java` with limited modules. Specifically `--limit-modules java.base,jdk.jfr`. For this particular test that is not sufficient to be able to run the test. `TestModularImage.java` uses the `ToolProvider` API to get a hold of `javac` as well as for `jlink` without them being present the test should be skipped. I've done that by adding those two required modules in the `@modules` tag. Testing: - [ ] GHA - [x] `jdk/jfr/jvm/TestModularImage.java` with test option `--limit-modules jdk.jfr,java.base` which now skips the test (instead of failing the test). Running `TestModularImage.java` without the `--limit-modules` option and it passes. Thoughts? ------------- Commit messages: - 8347496: Test jdk/jfr/jvm/TestModularImage.java fails after JDK-8347124: No javac Changes: https://git.openjdk.org/jdk/pull/23079/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=23079&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8347496 Stats: 1 line in 1 file changed: 1 ins; 0 del; 0 mod Patch: https://git.openjdk.org/jdk/pull/23079.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/23079/head:pull/23079 PR: https://git.openjdk.org/jdk/pull/23079

3 4

Re: RFR: 8347124: Clean tests with --enable-linkable-runtime [v3]
by Severin Gehwolf 13 Jan '25

13 Jan '25

> Please review this trivial test-only patch in support of running tests on JEP 493 enabled builds. Both tests use the `ToolProvider` API so as to run `jlink` in-process of the test JVM which includes module patches (as in - uses `--patch-module`) which is not supported for JEP 493 enabled JDKs. The proposed fix is to run those two tests in `/othervm` so as to no longer see the module patch of the test JDK. > > Testing: > - [x] GHA > - [x] Affected tests which fail prior the patch and pass after with a JEP 493 enabled JDK. Also tested on a JDK that includes `jmods` folder. > > Thoughts? Severin Gehwolf has updated the pull request incrementally with one additional commit since the last revision: Expand comment ------------- Changes: - all: https://git.openjdk.org/jdk/pull/22969/files - new: https://git.openjdk.org/jdk/pull/22969/files/6eceb0de..77dc9376 Webrevs: - full: https://webrevs.openjdk.org/?repo=jdk&pr=22969&range=02 - incr: https://webrevs.openjdk.org/?repo=jdk&pr=22969&range=01-02 Stats: 8 lines in 2 files changed: 6 ins; 0 del; 2 mod Patch: https://git.openjdk.org/jdk/pull/22969.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/22969/head:pull/22969 PR: https://git.openjdk.org/jdk/pull/22969

3 3

RFR: 8343510: JFR: Remove AccessControlContext from FlightRecorder::addListener specification
by Erik Gahlin 10 Jan '25

10 Jan '25

Could I have a review that updates the specification for FlightRecorder::addListener to reflect that the Security Manager is now disabled. Testing: jdk/jdk/jfr Thanks Erik ------------- Commit messages: - Initial Changes: https://git.openjdk.org/jdk/pull/23006/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=23006&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8343510 Stats: 2 lines in 1 file changed: 0 ins; 1 del; 1 mod Patch: https://git.openjdk.org/jdk/pull/23006.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/23006/head:pull/23006 PR: https://git.openjdk.org/jdk/pull/23006

2 2

[jdk24] RFR: 8345337: JFR: jfr view should display all direct subfields for an event type
by Erik Gahlin 10 Jan '25

10 Jan '25

8345337: JFR: jfr view should display all direct subfields for an event type ------------- Commit messages: - Backport 672c413c61d9b155020a0fd4bd1c2bc0661a60fb Changes: https://git.openjdk.org/jdk/pull/23009/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=23009&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8345337 Stats: 4 lines in 1 file changed: 1 ins; 0 del; 3 mod Patch: https://git.openjdk.org/jdk/pull/23009.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/23009/head:pull/23009 PR: https://git.openjdk.org/jdk/pull/23009

2 2

RFR: 8347287: JFR: Remove use of Security Manager
by Erik Gahlin 10 Jan '25

10 Jan '25

Could I have a review of PR that removes Java related code not needed when the Security Manager is disabled. Testing: tier1 + test/jdk/jdk/jfr Thanks Erik ------------- Commit messages: - Update copyright - Update specification in seperate issue - Remove mistakenly added directory - Fix copyright year - Remove whitespace - Remove whitespace - Update copyright to 2025 - Initial Changes: https://git.openjdk.org/jdk/pull/22787/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=22787&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8347287 Stats: 1986 lines in 86 files changed: 197 ins; 1380 del; 409 mod Patch: https://git.openjdk.org/jdk/pull/22787.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/22787/head:pull/22787 PR: https://git.openjdk.org/jdk/pull/22787

3 5

Integrated: 8347124: Clean tests with --enable-linkable-runtime
by Severin Gehwolf 10 Jan '25

10 Jan '25

On Wed, 8 Jan 2025 14:24:25 GMT, Severin Gehwolf <sgehwolf(a)openjdk.org> wrote: > Please review this trivial test-only patch in support of running tests on JEP 493 enabled builds. Both tests use the `ToolProvider` API so as to run `jlink` in-process of the test JVM which includes module patches (as in - uses `--patch-module`) which is not supported for JEP 493 enabled JDKs. The proposed fix is to run those two tests in `/othervm` so as to no longer see the module patch of the test JDK. > > Testing: > - [x] GHA > - [x] Affected tests which fail prior the patch and pass after with a JEP 493 enabled JDK. Also tested on a JDK that includes `jmods` folder. > > Thoughts? This pull request has now been integrated. Changeset: 1f457977 Author: Severin Gehwolf <sgehwolf(a)openjdk.org> URL: https://git.openjdk.org/jdk/commit/1f457977f062e4ed219c6fa0fe26cb42acaf4bf2 Stats: 14 lines in 2 files changed: 10 ins; 0 del; 4 mod 8347124: Clean tests with --enable-linkable-runtime Reviewed-by: shade, alanb ------------- PR: https://git.openjdk.org/jdk/pull/22969

1 0

Socket Read event always reads 5B on jdk 23.0.1+11
by Steven Schlansker 09 Jan '25

09 Jan '25

Hi hotspot-jfr-dev, I am trying to track down a problem with long db queries - Java reports 10s, but the database server says the query only took 100ms. In the flight recording, I look up the Socket Read events and indeed find my suspiciously long read event. However, the event reports that only 5B are read, which is hard to believe. In fact, *every* socket read event reports exactly 5B read. This seems impossible - the server builds a 45MB response body, and all the data comes from the database. Should I expect that the socket read event reliably reports Bytes Read? I don't believe this socket stream uses asynchronous IO, which doesn't seem to be recorded at all (bummer!) I tried to read the Socket$SocketInputStream Java code but the instrumentation seems to happen due to much deeper magic. We run openjdk 23.0.1+11 on rockylinux 9.5 Example event log, subset: Start Time Duration End Time Event Thread Remote Address Bytes Read End of Stream Remote Host Remote Port Timeout Value Event Type Id 1734454183069711840 epochns 24210111038 ticks 3815799226963477504 epochticks company-server-56 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead 1734454183748414110 epochns 20222774822 ticks 3815799224469286400 epochticks company-server-39 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead 1734454590578194253 epochns 4759793838 ticks 3815800104031821312 epochticks company-server-59 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead 1734454441187707733 epochns 4751996960 ticks 3815799775364954112 epochticks company-server-42 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead 1734454425227497984 epochns 4120098692 ticks 3815799739620594688 epochticks company-server-39 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead 1734454241790648566 epochns 1079527886 ticks 3815799333018955264 epochticks company-server-55 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead 1734454590289810962 epochns 884150534 ticks 3815800099521735168 epochticks company-server-47 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead 1734454649681226560 epochns 757929740 ticks 3815800230056628736 epochticks company-server-47 10.28.224.20 5 B false wsdb 5432 0 ms jdk.SocketRead An example stack trace for one of the events: void jdk.internal.event.SocketReadEvent.emit(long, long, long, SocketAddress, long) int java.net.Socket$SocketInputStream.read(byte[], int, int) int sun.security.ssl.SSLSocketInputRecord.read(InputStream, byte[], int, int) int sun.security.ssl.SSLSocketInputRecord.readHeader() int sun.security.ssl.SSLSocketInputRecord.bytesInCompletePacket() ByteBuffer sun.security.ssl.SSLSocketImpl.readApplicationRecord(ByteBuffer) int sun.security.ssl.SSLSocketImpl$AppInputStream.read(byte[], int, int) boolean org.postgresql.core.VisibleBufferedInputStream.readMore(int, boolean) boolean org.postgresql.core.VisibleBufferedInputStream.ensureBytes(int, boolean) boolean org.postgresql.core.VisibleBufferedInputStream.ensureBytes(int) int org.postgresql.core.VisibleBufferedInputStream.read() int org.postgresql.core.PGStream.receiveChar() void org.postgresql.core.v3.QueryExecutorImpl.processResults(ResultHandler, int, boolean) void org.postgresql.core.v3.QueryExecutorImpl.execute(Query, ParameterList, ResultHandler, int, int, int, boolean) void org.postgresql.jdbc.PgStatement.executeInternal(CachedQuery, ParameterList, int) void org.postgresql.jdbc.PgStatement.execute(CachedQuery, ParameterList, int) boolean org.postgresql.jdbc.PgPreparedStatement.executeWithFlags(int) boolean org.postgresql.jdbc.PgPreparedStatement.execute() boolean com.zaxxer.hikari.pool.ProxyPreparedStatement.execute() boolean com.zaxxer.hikari.pool.HikariProxyPreparedStatement.execute() Thank you for any insight - happy to provide more information if possible.

2 1

Re: RFR: 8347124: Clean tests with --enable-linkable-runtime [v2]
by Severin Gehwolf 09 Jan '25

09 Jan '25

> Please review this trivial test-only patch in support of running tests on JEP 493 enabled builds. Both tests use the `ToolProvider` API so as to run `jlink` in-process of the test JVM which includes module patches (as in - uses `--patch-module`) which is not supported for JEP 493 enabled JDKs. The proposed fix is to run those two tests in `/othervm` so as to no longer see the module patch of the test JDK. > > Testing: > - [ ] GHA > - [x] Affected tests which fail prior the patch and pass after with a JEP 493 enabled JDK. Also tested on a JDK that includes `jmods` folder. > > Thoughts? Severin Gehwolf has updated the pull request incrementally with one additional commit since the last revision: Copyright and @comment ------------- Changes: - all: https://git.openjdk.org/jdk/pull/22969/files - new: https://git.openjdk.org/jdk/pull/22969/files/afc33798..6eceb0de Webrevs: - full: https://webrevs.openjdk.org/?repo=jdk&pr=22969&range=01 - incr: https://webrevs.openjdk.org/?repo=jdk&pr=22969&range=00-01 Stats: 6 lines in 2 files changed: 4 ins; 0 del; 2 mod Patch: https://git.openjdk.org/jdk/pull/22969.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/22969/head:pull/22969 PR: https://git.openjdk.org/jdk/pull/22969

3 5