fix(codecache): report the real runtime-stubs cache size#681
Open
rkennke wants to merge 5 commits into
Open
Conversation
Contributor
There was a problem hiding this comment.
Pull request overview
Fixes the CODECACHE_RUNTIME_STUBS_SIZE_BYTES counter to report the HotSpot JIT runtime-stubs cache size (instead of incorrectly duplicating the native-libraries code cache size), improving the accuracy of codecache-related diagnostics emitted by Profiler::dump().
Changes:
- Wire
CODECACHE_RUNTIME_STUBS_SIZE_BYTEStoJitCodeCache::_runtime_stubs.memoryUsage()instead ofnative_libs.memoryUsage(). - Add
JitCodeCache::runtimeStubsMemoryUsage()that reads under the shared stubs lock to avoid races with concurrent JVMTIDynamicCodeGenerated()updates. - Include the necessary
hotspot/jitCodeCache.hdependency fromprofiler.cpp.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| ddprof-lib/src/main/cpp/profiler.cpp | Sets CODECACHE_RUNTIME_STUBS_SIZE_BYTES from JitCodeCache::runtimeStubsMemoryUsage() instead of duplicating native libs size. |
| ddprof-lib/src/main/cpp/hotspot/jitCodeCache.h | Declares runtimeStubsMemoryUsage() with concurrency rationale. |
| ddprof-lib/src/main/cpp/hotspot/jitCodeCache.cpp | Implements runtimeStubsMemoryUsage() using _stubs_lock.lockShared() around _runtime_stubs.memoryUsage(). |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Contributor
CI Test ResultsRun: #30019112376 | Commit:
Status Overview
Legend: ✅ passed | ❌ failed | ⚪ skipped | 🚫 cancelled Failed Testsglibc-aarch64/debug / 25-graalJob: View logs No detailed failure information available. Check the job logs. Summary: Total: 32 | Passed: 31 | Failed: 1 Updated: 2026-07-23 15:35:52 UTC |
Contributor
Benchmark Results (commit 8c422a3)Pipeline: https://gitlab.ddbuild.io/DataDog/apm-reliability/benchmarking-platform/-/pipelines/126245147 Commit: ✅ Within expected boundariesNo significant runtime deltas (all within run-to-run noise) and no internal-counter outliers. Runtime details (per benchmark × JDK)
Internal counter details (ddprof)ddprof internal counters, latest / dev (✅ = 0, · = unavailable):
|
rkennke
marked this pull request as ready for review
July 22, 2026 14:21
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 8c422a3043
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
Contributor
Reliability & Chaos Results✅ All reliability & chaos checks passed Pipeline: https://gitlab.ddbuild.io/DataDog/java-profiler/-/pipelines/126561737 |
Contributor
Author
|
Correct that the magnitude is still approximate here — but that's by design of the split, not something to fix in this PR:
So #681 + #677 together give an accurate runtime-stubs figure; keeping the accuracy fix in one place (#677) avoids duplicating/rebasing the same |
rkennke
force-pushed
the
fix/codecache-runtime-stubs-counter
branch
from
July 22, 2026 14:58
8c422a3 to
bb2147f
Compare
Contributor
Author
This comment has been minimized.
This comment has been minimized.
Contributor
Benchmark Results (commit 2886d41)Pipeline: https://gitlab.ddbuild.io/DataDog/apm-reliability/benchmarking-platform/-/pipelines/126276523 Commit:
|
| Benchmark | JDK | Latest | Dev | Δ (dev vs latest) | Issues L/D |
|---|---|---|---|---|---|
| akka-uct | 21 | ✅ 10301 ms (21 iters) | ✅ 10255 ms (21 iters) | ≈ -0.4% (±11.7%) | — / — |
| akka-uct | 25 | ✅ 8866 ms (24 iters) | ✅ 8851 ms (24 iters) | ≈ -0.2% (±10.2%) | — / — |
| finagle-chirper | 21 | ✅ 6010 ms (33 iters) | ✅ 5981 ms (33 iters) | ≈ -0.5% (±25%) | |
| finagle-chirper | 25 | ✅ 5450 ms (36 iters) | ✅ 5463 ms (36 iters) | ≈ +0.2% (±24.9%) | |
| fj-kmeans | 21 | ✅ 2695 ms (70 iters) | ✅ 2780 ms (67 iters) | 🔴 +3.2% | — / — |
| fj-kmeans | 25 | ✅ 2827 ms (66 iters) | ✅ 2809 ms (66 iters) | ≈ -0.6% (±2.6%) | — / — |
| future-genetic | 25 | ✅ 2115 ms (88 iters) | ✅ 2071 ms (90 iters) | ≈ -2.1% (±2.6%) | — / — |
| naive-bayes | 21 | ✅ 1322 ms (130 iters) | ✅ 1262 ms (136 iters) | ≈ -4.5% (±31.6%) | — / — |
| naive-bayes | 25 | ✅ 1020 ms (168 iters) | ✅ 999 ms (172 iters) | ≈ -2.1% (±31.3%) | — / — |
| reactors | 21 | ✅ 16568 ms (15 iters) | ✅ 17173 ms (15 iters) | ≈ +3.7% (±7.6%) | — / — |
| reactors | 25 | ✅ 18482 ms (15 iters) | ✅ 18271 ms (15 iters) | ≈ -1.1% (±5.5%) | — / — |
Internal counter details (ddprof)
ddprof internal counters, latest / dev (✅ = 0, · = unavailable):
| Benchmark | JDK | Dropped rec | Dropped jvmti | Dropped trace | Skipped WC | AGCT fail | Unwind fail |
|---|---|---|---|---|---|---|---|
| akka-uct | 21 | ✅ / ✅ | ✅ / ✅ | 2 / 1 | 1960 / 2005 | ✅ / ✅ | ✅ / ✅ |
| akka-uct | 25 | ✅ / ✅ | ✅ / ✅ | 3 / 3 | 2153 / 2334 | ✅ / ✅ | ✅ / ✅ |
| finagle-chirper | 21 | ✅ / ✅ | ✅ / ✅ | 2 / 1 | 8594 / 8571 | ✅ / ✅ | ✅ / ✅ |
| finagle-chirper | 25 | ✅ / ✅ | ✅ / ✅ | 1 / 1 | 8022 / 8344 | ✅ / ✅ | ✅ / ✅ |
| fj-kmeans | 21 | ✅ / ✅ | ✅ / ✅ | ✅ / 1 | 1273 / 1304 | ✅ / ✅ | ✅ / ✅ |
| fj-kmeans | 25 | ✅ / ✅ | ✅ / ✅ | 3 / ✅ | 1278 / 1268 | ✅ / ✅ | ✅ / ✅ |
| future-genetic | 25 | ✅ / ✅ | ✅ / ✅ | ✅ / 3 | 2864 / 2825 | ✅ / ✅ | ✅ / ✅ |
| naive-bayes | 21 | ✅ / ✅ | ✅ / ✅ | 4 / 2 | 3542 / 3487 | ✅ / ✅ | ✅ / ✅ |
| naive-bayes | 25 | ✅ / ✅ | ✅ / ✅ | 2 / ✅ | 3473 / 3500 | ✅ / ✅ | ✅ / ✅ |
| reactors | 21 | ✅ / ✅ | ✅ / ✅ | ✅ / ✅ | 1574 / 1749 | ✅ / ✅ | ✅ / ✅ |
| reactors | 25 | ✅ / ✅ | ✅ / ✅ | ✅ / 2 | 1873 / 1701 | ✅ / ✅ | ✅ / ✅ |
CODECACHE_RUNTIME_STUBS_SIZE_BYTES was set to native_libs.memoryUsage() — the same value as CODECACHE_NATIVE_SIZE_BYTES — a copy-paste since the counter was introduced (d0b0d1e). It never reflected the runtime-stubs cache. The runtime stubs live in JitCodeCache::_runtime_stubs (populated by DynamicCodeGenerated as the JVM emits stubs), a different cache from the native libraries. Add JitCodeCache::runtimeStubsMemoryUsage(), which reads that cache's size under its shared _stubs_lock (the cache is mutated concurrently by the JVMTI callback), and set the counter from it. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ootprint" CodeCache::memoryUsage() is a capacity/count-based estimate on main, so "live heap footprint" overstated it. Reword to "approximate heap usage" and note #677 makes memoryUsage() exact. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
rkennke
force-pushed
the
fix/codecache-runtime-stubs-counter
branch
from
July 23, 2026 10:52
2886d41 to
fd1ca88
Compare
Contributor
Benchmark Results (commit 0742b45)Pipeline: https://gitlab.ddbuild.io/DataDog/apm-reliability/benchmarking-platform/-/pipelines/126484118 Commit:
|
| Benchmark | JDK | Latest | Dev | Δ (dev vs latest) | Issues L/D |
|---|---|---|---|---|---|
| akka-uct | 21 | ✅ 10297 ms (21 iters) | ✅ 10304 ms (21 iters) | ≈ +0.1% (±11.3%) | — / — |
| akka-uct | 25 | ✅ 8896 ms (24 iters) | ✅ 8921 ms (24 iters) | ≈ +0.3% (±9.7%) | — / — |
| finagle-chirper | 21 | ✅ 5944 ms (33 iters) | ✅ 6013 ms (33 iters) | ≈ +1.2% (±25.9%) | |
| finagle-chirper | 25 | ✅ 5472 ms (36 iters) | ✅ 5453 ms (36 iters) | ≈ -0.3% (±24%) | |
| fj-kmeans | 21 | ✅ 2745 ms (68 iters) | ✅ 2819 ms (66 iters) | 🔴 +2.7% | — / — |
| fj-kmeans | 25 | ✅ 2835 ms (66 iters) | ✅ 2833 ms (66 iters) | ≈ -0.1% (±2.6%) | — / — |
| future-genetic | 21 | ✅ 2087 ms (89 iters) | ✅ 2041 ms (90 iters) | ≈ -2.2% (±2.6%) | — / — |
| future-genetic | 25 | ✅ 2096 ms (88 iters) | ✅ 2071 ms (89 iters) | ≈ -1.2% (±2.6%) | — / — |
| naive-bayes | 21 | ✅ 1243 ms (137 iters) | ✅ 1281 ms (134 iters) | ≈ +3.1% (±33.1%) | — / — |
| naive-bayes | 25 | ✅ 1000 ms (171 iters) | ✅ 1007 ms (170 iters) | ≈ +0.7% (±31.4%) | — / — |
| reactors | 21 | ✅ 16383 ms (15 iters) | ✅ 15786 ms (15 iters) | ≈ -3.6% (±7.5%) | — / — |
Internal counter details (ddprof)
ddprof internal counters, latest / dev (✅ = 0, · = unavailable):
| Benchmark | JDK | Dropped rec | Dropped jvmti | Dropped trace | Skipped WC | AGCT fail | Unwind fail |
|---|---|---|---|---|---|---|---|
| akka-uct | 21 | ✅ / ✅ | ✅ / ✅ | 3 / 5 | 2019 / 2035 | ✅ / ✅ | ✅ / ✅ |
| akka-uct | 25 | ✅ / ✅ | ✅ / ✅ | ✅ / ✅ | 2224 / 2122 | ✅ / ✅ | ✅ / ✅ |
| finagle-chirper | 21 | ✅ / ✅ | ✅ / ✅ | 2 / 3 | 8454 / 8884 | ✅ / ✅ | ✅ / ✅ |
| finagle-chirper | 25 | ✅ / ✅ | ✅ / ✅ | 2 / ✅ | 8262 / 8260 | ✅ / ✅ | ✅ / ✅ |
| fj-kmeans | 21 | ✅ / ✅ | ✅ / ✅ | 1 / 1 | 1253 / 1246 | ✅ / ✅ | ✅ / ✅ |
| fj-kmeans | 25 | ✅ / ✅ | ✅ / ✅ | ✅ / ✅ | 1279 / 1283 | ✅ / ✅ | ✅ / ✅ |
| future-genetic | 21 | ✅ / ✅ | ✅ / ✅ | 4 / 3 | 3028 / 2878 | ✅ / ✅ | ✅ / ✅ |
| future-genetic | 25 | ✅ / ✅ | ✅ / ✅ | ✅ / 2 | 2859 / 2838 | ✅ / ✅ | ✅ / ✅ |
| naive-bayes | 21 | ✅ / ✅ | ✅ / ✅ | 1 / 4 | 3497 / 3520 | ✅ / ✅ | ✅ / ✅ |
| naive-bayes | 25 | ✅ / ✅ | ✅ / ✅ | 3 / 2 | 3489 / 3438 | ✅ / ✅ | ✅ / ✅ |
| reactors | 21 | ✅ / ✅ | ✅ / ✅ | 2 / 1 | 1642 / 1682 | ✅ / ✅ | ✅ / ✅ |
jbachorik
reviewed
Jul 23, 2026
The runtime-stubs memory counter called the HotSpot-specific JitCodeCache::runtimeStubsMemoryUsage() directly from shared profiler.cpp, with no VM guard. On J9/Zing that reports a fixed near-zero baseline with nothing in the code marking the value as meaningless there. Route it through the same JVMSupport -> HotspotSupport abstraction the sibling isJitCode() accessor uses: JVMSupport::runtimeStubsMemoryUsage() dispatches to HotspotSupport (-> JitCodeCache) on HotSpot and returns 0 elsewhere. profiler.cpp no longer references JitCodeCache directly. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
Benchmark Results (commit b80dfc4)Pipeline: https://gitlab.ddbuild.io/DataDog/apm-reliability/benchmarking-platform/-/pipelines/126538514 Commit: ✅ Within expected boundariesNo significant runtime deltas (all within run-to-run noise) and no internal-counter outliers. Runtime details (per benchmark × JDK)
Internal counter details (ddprof)ddprof internal counters, latest / dev (✅ = 0, · = unavailable):
|
Contributor
Benchmark Results (commit f6ee1ce)Pipeline: https://gitlab.ddbuild.io/DataDog/apm-reliability/benchmarking-platform/-/pipelines/126561763 Commit:
|
| Benchmark | JDK | Latest | Dev | Δ (dev vs latest) | Issues L/D |
|---|---|---|---|---|---|
| akka-uct | 21 | ✅ 10224 ms (21 iters) | ✅ 10278 ms (21 iters) | ≈ +0.5% (±12.1%) | — / — |
| akka-uct | 25 | ✅ 8834 ms (24 iters) | ✅ 8870 ms (24 iters) | ≈ +0.4% (±10%) | — / — |
| finagle-chirper | 21 | ✅ 6008 ms (33 iters) | ✅ 6014 ms (33 iters) | ≈ +0.1% (±24.7%) | |
| fj-kmeans | 21 | ✅ 2771 ms (68 iters) | ✅ 2774 ms (68 iters) | ≈ +0.1% (±2.8%) | — / — |
| fj-kmeans | 25 | ✅ 2803 ms (66 iters) | ✅ 2837 ms (66 iters) | ≈ +1.2% (±2.6%) | — / — |
| future-genetic | 21 | ✅ 2071 ms (89 iters) | ✅ 2162 ms (86 iters) | 🔴 +4.4% | — / — |
| naive-bayes | 21 | ✅ 1241 ms (139 iters) | ✅ 1297 ms (133 iters) | ≈ +4.5% (±33%) | — / — |
| naive-bayes | 25 | ✅ 970 ms (176 iters) | ✅ 997 ms (171 iters) | ≈ +2.8% (±32%) | — / — |
| reactors | 21 | ✅ 15680 ms (15 iters) | ✅ 16453 ms (15 iters) | ≈ +4.9% (±7.7%) | — / — |
| reactors | 25 | ✅ 18106 ms (15 iters) | ✅ 18265 ms (15 iters) | ≈ +0.9% (±5.8%) | — / — |
Internal counter details (ddprof)
ddprof internal counters, latest / dev (✅ = 0, · = unavailable):
| Benchmark | JDK | Dropped rec | Dropped jvmti | Dropped trace | Skipped WC | AGCT fail | Unwind fail |
|---|---|---|---|---|---|---|---|
| akka-uct | 21 | ✅ / ✅ | ✅ / ✅ | ✅ / ✅ | ✅ / ✅ | ✅ / ✅ | ✅ / ✅ |
| akka-uct | 25 | ✅ / ✅ | ✅ / ✅ | 3 / 2 | 2339 / 2394 | ✅ / ✅ | ✅ / ✅ |
| finagle-chirper | 21 | ✅ / ✅ | ✅ / ✅ | 2 / 5 | 8689 / 8640 | ✅ / ✅ | ✅ / ✅ |
| fj-kmeans | 21 | ✅ / ✅ | ✅ / ✅ | 3 / ✅ | 1288 / 1264 | ✅ / ✅ | ✅ / ✅ |
| fj-kmeans | 25 | ✅ / ✅ | ✅ / ✅ | 4 / 1 | 1254 / 1290 | ✅ / ✅ | ✅ / ✅ |
| future-genetic | 21 | ✅ / ✅ | ✅ / ✅ | 1 / ✅ | 3092 / 2993 | ✅ / ✅ | ✅ / ✅ |
| naive-bayes | 21 | ✅ / ✅ | ✅ / ✅ | 8 / 1 | 3576 / 3480 | ✅ / ✅ | ✅ / ✅ |
| naive-bayes | 25 | ✅ / ✅ | ✅ / ✅ | 6 / 2 | 3504 / 3472 | ✅ / ✅ | ✅ / ✅ |
| reactors | 21 | ✅ / ✅ | ✅ / ✅ | 1 / 1 | 1552 / 1675 | ✅ / ✅ | ✅ / ✅ |
| reactors | 25 | ✅ / ✅ | ✅ / ✅ | 4 / 2 | 1814 / 1744 | ✅ / ✅ | ✅ / ✅ |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Fixes
CODECACHE_RUNTIME_STUBS_SIZE_BYTESto report the JIT runtime-stubs code cache, instead of duplicating the native-libraries size.Why
The counter was set to
native_libs.memoryUsage()— the same expression asCODECACHE_NATIVE_SIZE_BYTESon the line above — a copy-paste present since it was introduced (d0b0d1e44 "add counters for codecaches"). So it never reflected the runtime stubs; it just double-reported the native-libs total.The runtime stubs are a separate cache:
JitCodeCache::_runtime_stubs(name"[stubs]"), populated viaJitCodeCache::DynamicCodeGenerated()→add()as the JVM emits VM stubs. (There is also an unrelated, unusedLibraries::_runtime_stubsmember — same name, different thing; the live one is inJitCodeCache.)How
JitCodeCache::runtimeStubsMemoryUsage(), which reads_runtime_stubs.memoryUsage()under_stubs_lock.lockShared(). That cache is mutated concurrently by theDynamicCodeGeneratedJVMTI callback (add()/expand()under_stubs_lock), so the read must hold the shared lock — mirroringfindRuntimeStub().CODECACHE_RUNTIME_STUBS_SIZE_BYTESfrom it inProfiler::dump().hotspot/jitCodeCache.*is compiled for all VM targets; on non-HotSpot VMs the callback never fires, so the cache is empty and the counter reports its (small, constant) empty-cache baseline rather than the wrong native-libs figure.Notes
CodeCache::memoryUsage()is accurate, so this counter reports the real runtime-stubs footprint.main(which includes the merged fix(codecache): make memoryUsage() accurate and live #677 accuracy fix), so no duplicate accuracy work is needed here._runtime_stubsis private and populated only by the JVMTI callback, so it can't be driven from gtest. Exercised by the JVM integration tests in CI (dump populates the counter).Testing
:ddprof-lib:gtestDebuggreen — 349 tests, 0 failures.🤖 Generated with Claude Code