January 2024 - hotspot-jfr-dev

RFR: 8323540: assert((!((((method)->is_trace_flag_set(((1 << 4) << 8))))))) failed: invariant
by Markus Grönlund 10 Jan '24

10 Jan '24

The recent integration of [JDK-8316241](https://bugs.openjdk.org/browse/JDK-8316241) is missing a commit. This is that missing commit. Thanks Markus ------------- Commit messages: - 8323540 Changes: https://git.openjdk.org/jdk/pull/17351/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=17351&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8323540 Stats: 1 line in 1 file changed: 0 ins; 1 del; 0 mod Patch: https://git.openjdk.org/jdk/pull/17351.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/17351/head:pull/17351 PR: https://git.openjdk.org/jdk/pull/17351

2 2

[jdk22] RFR: 8316241: Test jdk/jdk/jfr/jvm/TestChunkIntegrity.java failed
by Markus Grönlund 10 Jan '24

10 Jan '24

8316241: Test jdk/jdk/jfr/jvm/TestChunkIntegrity.java failed ------------- Commit messages: - Backport b2a39c576706622b624314c89fa6d10d0b422f86 Changes: https://git.openjdk.org/jdk22/pull/53/files Webrev: https://webrevs.openjdk.org/?repo=jdk22&pr=53&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8316241 Stats: 949 lines in 5 files changed: 418 ins; 323 del; 208 mod Patch: https://git.openjdk.org/jdk22/pull/53.diff Fetch: git fetch https://git.openjdk.org/jdk22.git pull/53/head:pull/53 PR: https://git.openjdk.org/jdk22/pull/53

2 2

RFR: 8316241: Test jdk/jdk/jfr/jvm/TestChunkIntegrity.java failed
by Markus Grönlund 10 Jan '24

10 Jan '24

Greetings, This fix allows JFR artefact tagging to become deep instead of insufficiently shallow. It adjusts a race condition where one thread tags a class as being in use while another thread tags a class to be both method and used. Testing: jdk_jfr, stress_testing Thanks Markus ------------- Commit messages: - 8316241 Changes: https://git.openjdk.org/jdk/pull/17328/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=17328&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8316241 Stats: 949 lines in 5 files changed: 418 ins; 323 del; 208 mod Patch: https://git.openjdk.org/jdk/pull/17328.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/17328/head:pull/17328 PR: https://git.openjdk.org/jdk/pull/17328

2 2

RFR: 8321698: JFR asserts in write_classloader: failed invariant
by Markus Grönlund 09 Jan '24

09 Jan '24

Greetings, I have reviewed the JFR tagging mechanism, which lacks proper transitivity in certain situations. For example, when tagging a PackageEntry or a ModuleEntry, no tag information is set for the ModuleEntry's CLD and its class_loader_klass. The assert then fires when writing the CLD of a ModuleEntry, should its CLD not have been tagged by some other means (some other direct usage). The tagging must, in general, become deeper than what is done today, i.e. it must cover the transitive closure of a Klass, which involves two branches: 1) Klass -> CLD -> class_loader_klass 2) Klass -> PkgEntry -> ModuleEntry -> CLD -> class_loader_klass A Klass under its closure can be treated as a node in a binary tree. Its closure (subtree) is covered by writing and tagging of artifacts using preorder traversal by recursing its CLD klasses. Testing: jdk_jfr, stress testing Thanks Markus ------------- Commit messages: - 8321698 Changes: https://git.openjdk.org/jdk/pull/17177/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=17177&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8321698 Stats: 928 lines in 4 files changed: 395 ins; 315 del; 218 mod Patch: https://git.openjdk.org/jdk/pull/17177.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/17177/head:pull/17177 PR: https://git.openjdk.org/jdk/pull/17177

2 4

[jdk22] RFR: 8322142: JFR: Periodic tasks aren't orphaned between recordings
by Erik Gahlin 08 Jan '24

08 Jan '24

8322142: JFR: Periodic tasks aren't orphaned between recordings ------------- Commit messages: - Backport 1551928502c8ed96350e7b4f1316ea35587407fe Changes: https://git.openjdk.org/jdk22/pull/31/files Webrev: https://webrevs.openjdk.org/?repo=jdk22&pr=31&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8322142 Stats: 7 lines in 1 file changed: 0 ins; 4 del; 3 mod Patch: https://git.openjdk.org/jdk22/pull/31.diff Fetch: git fetch https://git.openjdk.org/jdk22.git pull/31/head:pull/31 PR: https://git.openjdk.org/jdk22/pull/31

2 2

[jdk22] RFR: 8322489: 22-b27: Up to 7% regression in all Footprint3-*-G1/ZGC
by Markus Grönlund 08 Jan '24

08 Jan '24

8322489: 22-b27: Up to 7% regression in all Footprint3-*-G1/ZGC ------------- Commit messages: - Backport 09c6c4ff021b7dc719c0b1e0dfb041b03bba1b5d Changes: https://git.openjdk.org/jdk22/pull/39/files Webrev: https://webrevs.openjdk.org/?repo=jdk22&pr=39&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8322489 Stats: 32 lines in 6 files changed: 19 ins; 3 del; 10 mod Patch: https://git.openjdk.org/jdk22/pull/39.diff Fetch: git fetch https://git.openjdk.org/jdk22.git pull/39/head:pull/39 PR: https://git.openjdk.org/jdk22/pull/39

2 2

RFR: 8322489 : 22-b27: Up to 7% regression in all Footprint3-*-G1/ZGC
by Markus Grönlund 08 Jan '24

08 Jan '24

Greetings, A footprint regression was accidentally introduced with [JDK-8211238](https://bugs.openjdk.org/browse/JDK-8211238). A small part of the JFR system had to be initialized early to support the functionality. We later decided that we only support this functionality iff the system is started with -XX:StartFlightRecording on the command-line. This PR will adjust to only perform early initialization under this premise. Testing: jdk_jfr Thanks Markus ------------- Commit messages: - 8322489 Changes: https://git.openjdk.org/jdk/pull/17179/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=17179&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8322489 Stats: 47 lines in 5 files changed: 29 ins; 9 del; 9 mod Patch: https://git.openjdk.org/jdk/pull/17179.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/17179/head:pull/17179 PR: https://git.openjdk.org/jdk/pull/17179

2 4

RFR: 8322142: JFR Periodic tasks aren't orphaned between recordings
by Carter Kozak 03 Jan '24

03 Jan '24

This issue is described on the jfr-dev mailing list here: https://mail.openjdk.org/pipermail/hotspot-jfr-dev/2023-December/005697.html With a minimal reproducer here: https://github.com/carterkozak/periodic-event-repro Currently a draft because I don't have a ticket for this issue, and I haven't had a chance to add a test for this. I have verified that a build with this change resolves the bug I encountered. ------------- Commit messages: - Update based on review from egahlin - Allow batch re-insertion after it has been orphaned - JFR Periodic tasks aren't orphaned between recordings Changes: https://git.openjdk.org/jdk/pull/17114/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=17114&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8322142 Stats: 7 lines in 1 file changed: 0 ins; 4 del; 3 mod Patch: https://git.openjdk.org/jdk/pull/17114.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/17114/head:pull/17114 PR: https://git.openjdk.org/jdk/pull/17114

2 8