Daily PR Report — 2026-06-11

rocprofiler-systems · ROCm/rocm-systems · label: project: rocprofiler-systems

Executive Summary

Total PRs: 46
In Review: 8 non-draft PRs with passing CI awaiting or with review decision
CI Failures: 14 PRs with at least one failing check
Missing Tests: 14 PRs with source changes but no test updates
Stale: 2 PRs with no activity in 10+ days (non-draft)

This is the first report — no overnight delta available.

Key Highlights

Ready to merge — action needed: Three approved PRs with passing CI have been sitting unmerged: rocm-systems#6460 (rocprofsys_push_trace_with_args, approved by dgaliffiAMD, 14 days old), rocm-systems#5819 (per-link PMC/track name alignment, approved by mradosav-amd, 35 days old, carries a Stale label), and rocm-systems#5414 (rocprof-sys-attach pytest, approved by two reviewers, 47 days old). These should be merged or explicitly held with a reason.

Long-running PRs accumulating risk: rocm-systems#4993 (HIP graph tests, 58 days old, approved, CI failing) and rocm-systems#5267 (rocpd API versioning, 50 days old, changes requested by jrmadsen, CI failing) are among the oldest non-draft PRs and both have unresolved CI failures — these need explicit prioritization or triage.

Major backend migration PRs are interconnected and fragile: rocm-systems#5808 (replace SQLite3/rocpd with rocprofiler-hub, 35 days old, CI failing on build and system-deps), rocm-systems#5267 (rocpd versioning), and rocm-systems#6528 (skip missing agents) all touch the rocpd_processor stack. Changes-requested on #6528 plus CI failures across the group suggest these need coordinated review.

C++20 modernization series has persistent CI failures and no tests: rocm-systems#5992 (replace SFINAE with C++20 concepts, 30 days old, 21 source files, 0 test files) and its follow-up rocm-systems#6803 (SFINAE/traits sweep, 6 days stale, 29 source files, 0 test files) both have failing CI and no test coverage. rocm-systems#5995 (jthread, 30 days old, also no tests) is in the same family.

High draft volume signals ongoing large efforts: 20 draft PRs are open, several of which are substantial (e.g., rocm-systems#6006 touches 67 source files; rocm-systems#5254 touches 89 source files). These are not urgent but warrant periodic review to prevent further accumulation.

PR Breakdown

Needs Attention (14)

PRs with failing CI or changes requested.

PR	Title	Author	CI	Review Status	Failing Checks / Blocker	Last Updated	Age
rocm-systems#6978	Improve output format handling in cli arguments	@marantic-amd	❌	Pending	TheRock CI Summary; Test rocprofiler-systems shard 1	2026-06-11	1d
rocm-systems#6803	extend C++20 concept conversions (SFINAE/traits sweep)	@adjordje-amd	❌	Unknown	TheRock CI Summary; Test rocprofiler-compute shard 2	2026-06-04	6d
rocm-systems#6735	fix(unified-memory): Cover missing KFD events in default run	@anujshuk-amd	❌	Pending	TheRock CI Summary; Test rocprofiler-sdk shard 1; ubuntu-noble-sanitizers undefined	2026-06-10	7d
rocm-systems#6647	Respect per-target LLVM directory in search paths	@jhuber6	❌	Pending	TheRock CI Summary; all 4 test suites failing; code-coverage	2026-06-02	9d
rocm-systems#6528	Fix: skip records with missing agents instead of aborting	@anujshuk-amd	✅	⚠️ Changes requested by adjordje-amd	Changes requested · no test files	2026-06-03	13d
rocm-systems#6461	Add test filter standardization to rocprofiler-systems	@dileepr1	❌	⚠️ Changes requested by kcossett-amd	rocprofiler-systems-release rhel 8.10 / opensuse 15.6	2026-06-10	14d
rocm-systems#6243	Switch Perfetto processing in cached data to run in parallel	@marantic-amd	❌	Pending	TheRock CI Summary; Test rocprofiler-systems shard 1	2026-06-10	22d
rocm-systems#6237	Improve Generated Output summary. Cleanup of noisy logs	@marantic-amd	❌	Pending	TheRock CI Summary; Build Linux Packages	2026-06-08	22d
rocm-systems#5995	trace_cache/buffer_storage: replace flush thread with std::jthread	@adjordje-amd	❌	Pending	TheRock CI Summary; rhel g++ 9 7.2; Test rocprofiler-compute shard 2	2026-06-10	30d
rocm-systems#5992	replace SFINAE traits with C++20 concepts	@adjordje-amd	❌	Unknown	TheRock CI Summary; Test rocprofiler-compute shard 2; Test rocprofiler-systems shard 1	2026-06-04	30d
rocm-systems#5867	Add OpenMP profiling doc	@kcossett-amd	✅	⚠️ Changes requested by prbasyal-amd (2 approvals: lstanisi-amd, dgaliffiAMD)	Changes requested by prbasyal-amd	2026-06-11	34d
rocm-systems#5808	Replace SQLite3/rocpd backend with rocprofiler-hub library	@anujshuk-amd	❌	Pending	TheRock CI Summary; Build Linux Packages; ubuntu-jammy/noble amdclang++ system-deps	2026-06-08	35d
rocm-systems#5267	Update rocprofiler-sdk-rocpd api to support versioning	@yhuiYH	❌	⚠️ Changes requested by jrmadsen	Code Coverage mi325 ubuntu-22.04	2026-06-10	50d
rocm-systems#4993	Add hip graph tests	@sputhala-amd	❌	✅ Approved by kcossett-amd	TheRock CI Summary; Test rocprofiler-systems/sdk/compute shard failures; ubuntu-noble g++ 6.3	2026-06-08	58d

In Review (8)

Passing CI, approved or awaiting review.

PR	Title	Author	CI	Review Status	Last Updated	Age
rocm-systems#7022	docs(rocprof-sys): add unified memory profiling how-to	@habajpai-amd	✅	Pending	2026-06-11	1d
rocm-systems#6992	Allow object level filtering	@kcossett-amd	✅	Pending	2026-06-10	1d
rocm-systems#6835	[rdc] CPX partition mode support + fixed-window profiler eval fields	@brichang	✅	✅ Approved by dmitrii-galantsev	2026-06-11	6d
rocm-systems#6535	Handle missing non-main thread orphaned OMPT+libomp events	@kcossett-amd	✅	Pending	2026-06-09	13d
rocm-systems#6460	Add `rocprofsys_push_trace_with_args`	@kcossett-amd	✅	✅ Approved by dgaliffiAMD	2026-06-11	14d
rocm-systems#6300	feat(rocprof-sys): add SPM beta scaffolding	@habajpai-amd	✅	Pending	2026-06-10	21d
rocm-systems#5819	align per-link PMC + track names across emit and register sites	@adjordje-amd	✅	✅ Approved by mradosav-amd	2026-06-09	35d
rocm-systems#5414	Add pytest for rocprof-sys-attach (attach/detach/re-attach)	@adjordje-amd	✅	✅ Approved by kcossett-amd, dgaliffiAMD	2026-06-10	47d

In Progress (2)

Active work, CI running.

PR	Title	Author	CI	Review Status	Last Updated	Age
rocm-systems#7074	Bump rocm-docs-core 1.33.1→1.35.0 (dependabot)	[@dependabot[bot]](https://github.com/apps/dependabot)	✅	Unknown	2026-06-11	0d
rocm-systems#6906	Remove dependency on Timemory units	@lloginov-amd	⏳	✅ Approved by mradosav-amd	2026-06-11	2d

Drafts (20)

PR	Title	Author	CI	Age
rocm-systems#7089	Rename driver to backend	@mradosav-amd	⏳	0d
rocm-systems#7087	test(rocprof-sys-sample): add data-driven CLI help coverage	@anujshuk-amd	⏳	0d
rocm-systems#7068	Fix AI NIC Phase 2 missing tracks and add diagnostics	@ajanicijamd	❌	0d
rocm-systems#7015	Change SMI temperature API to amdsmi_get_temp_metric	@sputhala-amd	❌	1d
rocm-systems#6989	Remove dependency on Timemory string manipulation	@lloginov-amd	✅	1d
rocm-systems#6892	Add JSON output support to rocprof-sys-avail	@sputhala-amd	❌	3d
rocm-systems#6891	Add live AMD SMI metric availability to rocprof-sys-avail	@sputhala-amd	❌	3d
rocm-systems#6652	Clean up pytest code	@kcossett-amd	❌	9d
rocm-systems#6532	[DRAFT] Testing multiarch tarballs	@jbonnell-amd	❌	13d
rocm-systems#6297	refactor(rocprof-sys-causal): centralize runtime state	@habajpai-amd	✅	21d
rocm-systems#6293	RCCL: Enable RMA CE path	@amd-arozanov	✅	22d
rocm-systems#6254	Config Robustness: Negative Tests for Invalid Values	@habajpai-amd	✅	22d
rocm-systems#6240	Unify control logic with different triggers	@marantic-amd	❌	22d
rocm-systems#6116	Add `rocprofsys_push_trace_with_args` and cache region arguments	@kcossett-amd	✅	27d
rocm-systems#6057	Wall Clock migration from timemory	@sputhala-amd	✅	29d
rocm-systems#6006	rocprofsys exception hierarchy + clean stacktrace	@adjordje-amd	❌	30d
rocm-systems#5762	Add RHEL system deps workflow	@sputhala-amd	❌	36d
rocm-systems#5285	Argparser refactor	@marantic-amd	✅	49d
rocm-systems#5254	Add clang-tidy in formatting workflow	@sputhala-amd	❌	50d
rocm-systems#4769	Add initial strix halo changes for rocprofiler-systems CI	@jbonnell-amd	❌	64d

Risk Signals

CI Instability

PR	Title	Author	Failing Checks	Last Updated
rocm-systems#6978	Improve output format handling in cli arguments	@marantic-amd	TheRock CI Summary; Test rocprofiler-systems shard 1	2026-06-11
rocm-systems#6803	extend C++20 concept conversions (SFINAE/traits sweep)	@adjordje-amd	TheRock CI Summary; Test rocprofiler-compute shard 2	2026-06-04
rocm-systems#6735	fix(unified-memory): Cover missing KFD events in default run	@anujshuk-amd	TheRock CI Summary; Test rocprofiler-sdk shard 1; sanitizers undefined	2026-06-10
rocm-systems#6647	Respect per-target LLVM directory in search paths	@jhuber6	TheRock CI Summary; all 4 test shards; code-coverage	2026-06-02
rocm-systems#6461	Add test filter standardization to rocprofiler-systems	@dileepr1	rocprofiler-systems-release rhel 8.10; opensuse 15.6	2026-06-10
rocm-systems#6243	Switch Perfetto processing in cached data to run in parallel	@marantic-amd	TheRock CI Summary; Test rocprofiler-systems shard 1	2026-06-10
rocm-systems#6237	Improve Generated Output summary. Cleanup of noisy logs	@marantic-amd	TheRock CI Summary; Build Linux Packages	2026-06-08
rocm-systems#5995	trace_cache/buffer_storage: replace flush thread with std::jthread	@adjordje-amd	TheRock CI Summary; rhel g++ 9 7.2; Test rocprofiler-compute shard 2	2026-06-10
rocm-systems#5992	replace SFINAE traits with C++20 concepts	@adjordje-amd	TheRock CI Summary; Test rocprofiler-compute shard 2; Test rocprofiler-systems shard 1	2026-06-04
rocm-systems#5808	Replace SQLite3/rocpd backend with rocprofiler-hub library	@anujshuk-amd	TheRock CI Summary; Build Linux Packages; ubuntu amdclang++ system-deps	2026-06-08
rocm-systems#5267	Update rocprofiler-sdk-rocpd api to support versioning	@yhuiYH	Code Coverage mi325 ubuntu-22.04	2026-06-10
rocm-systems#4993	Add hip graph tests	@sputhala-amd	TheRock CI Summary; Test rocprofiler-systems/sdk/compute shards; ubuntu-noble g++ 6.3	2026-06-08

14/46 PRs have CI failures.

Test Gap Summary

24 PRs include test updates. 14 PRs are missing tests.

PR	Title	Author	Source Files Changed	Recommended Test Level	Risk of No Tests	Suggested Action
rocm-systems#6803	extend C++20 concept conversions (SFINAE/traits sweep)	@adjordje-amd	29	Unit	Medium	Add concept constraint tests alongside `test_traits.cpp` or extend existing type-trait unit tests
rocm-systems#6528	Fix: skip records with missing agents instead of aborting	@anujshuk-amd	3 (`agent_manager`, `rocpd_processor`)	Unit	High	Add tests verifying graceful skip behavior; likely adjacent to `test_rocpd_processor.cpp` (see PR #5808)
rocm-systems#5995	trace_cache/buffer_storage: replace flush thread with std::jthread	@adjordje-amd	2 (`buffer_storage.cpp/.hpp`)	Unit	High	Concurrency behavior change — add tests for flush/teardown ordering; extend `test_buffer_storage.cpp`
rocm-systems#5992	replace SFINAE traits with C++20 concepts	@adjordje-amd	21	Unit	Medium	Broad refactor across template machinery — add concept satisfaction tests; possible to reuse SFINAE test patterns
rocm-systems#6906	Remove dependency on Timemory units	@lloginov-amd	29	Unit / Integration	Medium	Unit migration — verify unit conversions with regression tests; check `units.hpp` for testable surface
rocm-systems#6992	Allow object level filtering	@kcossett-amd	4 (instrumentation CLI)	Integration	Medium	CLI filtering is user-visible behavior; add integration test exercising new filter flags
rocm-systems#6835	[rdc] CPX partition mode support + fixed-window profiler eval fields	@brichang	7 (RDC/SMI layer)	Integration	Medium	Hardware-facing feature; at minimum add unit tests for partition logic in `RdcPartitionImpl`
rocm-systems#5098	Fix output index calculation in transpose_a kernel	@jared-mcd-han	1 (`transpose.new.cpp`)	Unit	High	Bug fix with no tests — high risk of regression; add correctness test for transpose output indexing
rocm-systems#7015	Change SMI temperature API to amdsmi_get_temp_metric	@sputhala-amd	1 (`gpu_driver.hpp`)	Unit	Medium	API swap — verify new call path with mock SMI; extend GPU metric unit tests
rocm-systems#6891	Add live AMD SMI metric availability to rocprof-sys-avail	@sputhala-amd	9 (avail + SMI probe)	Integration	Medium	New feature surface; extend `test_binaries.py` to verify avail output includes SMI metrics
rocm-systems#6297	refactor(rocprof-sys-causal): centralize runtime state	@habajpai-amd	3 (causal runtime)	Unit	Medium	Refactor of runtime state — add state transition tests for causal profiler
rocm-systems#6293	RCCL: Enable RMA CE path	@amd-arozanov	1 (`therock_configure_ci.py`)	System	Low	CI config change — low risk, but verify CI matrix correctness by inspection
rocm-systems#5762	Add RHEL system deps workflow	@sputhala-amd	1 (`symbol.cpp`)	Unit	Low	Workflow/CI change; the `symbol.cpp` change should have a unit regression test
rocm-systems#6116	Add `rocprofsys_push_trace_with_args` and cache region arguments	@kcossett-amd	13 (instrument + API layer)	Integration	High	New public API — requires tests; note superseded by #6460 which has tests; confirm relationship

Stale PRs

Stale (25+ days no activity)

PR	Title	Author	Last Updated	Age	Recommendation