Refactor: converge A2A3 and A5 TMR internals - #1718
Merged
ChaoWao merged 1 commit intoAug 7, 2026
Merged
Conversation
|
Important Review skippedAuto incremental reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
📝 WalkthroughWalkthroughThe change aligns a2a3 and a5 runtime APIs, removes scheduler scan profiling, updates profiling documentation, replaces header guards with ChangesRuntime API and behavior
Estimated code review effort: 3 (Moderate) | ~20 minutes Possibly related issues
Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/a2a3/runtime/tensormap_and_ringbuffer/docs/profiling_levels.md`:
- Line 74: Update the scheduler output example in profiling_levels.md to remove
the stale scan row, matching the scheduler counter lists at the documented
sections. Ensure all scheduler output documentation consistently reflects only
counters that are emitted.
In `@src/a2a3/runtime/tensormap_and_ringbuffer/runtime/scheduler/pto_scheduler.h`:
- Line 388: Update the comment for init_data_from_layout() to state that
arena-internal pointer fields, including ready_queues[].slots, are bound by
wire_arena_pointers() rather than init_data_from_layout().
In `@src/a5/runtime/tensormap_and_ringbuffer/docs/profiling_levels.md`:
- Line 114: Remove the stale scan row from the Level 2 output example in
profiling_levels.md, leaving only the complete, dispatch, and idle phase
statistics to match the current emitted output.
🪄 Autofix
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: a50ecad5-36d4-45c3-928f-2c58d696ec92
📒 Files selected for processing (49)
src/a2a3/runtime/tensormap_and_ringbuffer/aicpu/aicpu_executor.cppsrc/a2a3/runtime/tensormap_and_ringbuffer/common/pto_runtime_status.hsrc/a2a3/runtime/tensormap_and_ringbuffer/docs/profiling_levels.mdsrc/a2a3/runtime/tensormap_and_ringbuffer/host/dep_gen_replay.hsrc/a2a3/runtime/tensormap_and_ringbuffer/orchestration/pto_orchestration_api.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/aicore_completion_mailbox.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/aicore_completion_mailbox_types.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_async_kernel_api.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_async_wait.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_completion_token.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_constants.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_dep_compute.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_orchestrator.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_ring_buffer.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_runtime2_types.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_tensormap.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/pto_types.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/runtime.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/scheduler/pto_scheduler.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/scheduler/scheduler_context.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/scheduler/scheduler_types.hsrc/a2a3/runtime/tensormap_and_ringbuffer/runtime/shared/pto_runtime2_init.cppsrc/a5/runtime/tensormap_and_ringbuffer/common/pto_runtime_status.hsrc/a5/runtime/tensormap_and_ringbuffer/docs/MULTI_RING.mdsrc/a5/runtime/tensormap_and_ringbuffer/docs/RUNTIME_LOGIC.mdsrc/a5/runtime/tensormap_and_ringbuffer/docs/profiling_levels.mdsrc/a5/runtime/tensormap_and_ringbuffer/host/dep_gen_replay.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/aicore_completion_mailbox.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/aicore_completion_mailbox_types.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/backend/sdma/sdma_completion_kernel.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/backend/sdma/sdma_completion_scheduler.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/backend/urma/urma_completion_kernel.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/backend/urma/urma_completion_scheduler.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_async_kernel_api.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_async_wait.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_completion_token.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_constants.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_dep_compute.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_orchestrator.cppsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_ring_buffer.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_runtime2_types.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_tensormap.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/pto_types.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/runtime.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/scheduler/pto_scheduler.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/scheduler/scheduler_cold_path.cppsrc/a5/runtime/tensormap_and_ringbuffer/runtime/scheduler/scheduler_context.hsrc/a5/runtime/tensormap_and_ringbuffer/runtime/scheduler/scheduler_types.htests/ut/cpp/a2a3/test_a2a3_fatal.cpp
yanghaoran29
force-pushed
the
codex/issue-1582-tmr-cleanup
branch
from
August 6, 2026 09:39
4bd115d to
4c1cebb
Compare
Align scope-mode forwarding, payload prefetch, scheduler layout, internal names, and profiling documentation across the TMR runtimes. Retain compatibility header guards in otherwise-unchanged headers so existing architecture detection remains available.
yanghaoran29
force-pushed
the
codex/issue-1582-tmr-cleanup
branch
from
August 6, 2026 11:38
4c1cebb to
7a6761e
Compare
4 tasks
yanghaoran29
added a commit
to yanghaoran29/simpler
that referenced
this pull request
Aug 10, 2026
Close the two P4/P5 residuals flagged in hw-native-sys#1582's 2026-08-09 re-audit (after hw-native-sys#1718 landed): R1 — a5 lacked the scope-guard regression test that a2a3 gained in hw-native-sys#1718. a5's `PTO2_SCOPE_GUARD(...)` has been variadic (and its MANUAL path live, read by `rt_scope_begin` -> `begin_scope(mode)`) since the hw-native-sys#1100 import, but `ut-a5` would not have caught a regression in MANUAL forwarding. Mirror `ScopeGuardForwardsModeUntilLexicalScopeExit` to `test_a5_fatal.cpp`: `FakeRuntime` gains `pending_scope_mode` (immediately after `ops`, the prefix the inline wrappers pun through) and `last_scope_mode`; `fake_scope_begin` records the mode; the test asserts AUTO then MANUAL forwarding across lexical scope exit. Both test files also extend the single `offsetof(ops) == 0` guard to a second one pinning `pending_scope_mode` at `sizeof(PTO2RuntimeOps *)`, since the `reinterpret_cast` now depends on the two-field prefix, not one. R2 — a2a3 `RUNTIME_LOGIC.md` still described the retained temp buffer as a single runner-owned instance, but the shared host-runtime callback `get_retained_temp_buffer(runner_ctx, pipeline_slot, ...)` is per-slot and a5's matching paragraph already says so. Align a2a3 to a5's wording ("each slot's buffer lives on the runner"). Also fix a stale accessor location in both trees' `MULTI_RING.md`: `PTO2TaskId`'s direct accessors live in `src/common/task_interface/pto_task_id.h`, not `pto_runtime2_types.h`. Verification: built and ran `test_a2a3_fatal` and `test_a5_fatal` — 4/4 tests pass each (including the new a5 ScopeGuard case).
yanghaoran29
added a commit
to yanghaoran29/simpler
that referenced
this pull request
Aug 10, 2026
Close the two P4/P5 residuals flagged in hw-native-sys#1582's 2026-08-09 re-audit (after hw-native-sys#1718 landed): R1 — a5 lacked the scope-guard regression test that a2a3 gained in hw-native-sys#1718. a5's `PTO2_SCOPE_GUARD(...)` has been variadic (and its MANUAL path live, read by `rt_scope_begin` -> `begin_scope(mode)`) since the hw-native-sys#1100 import, but `ut-a5` would not have caught a regression in MANUAL forwarding. Mirror `ScopeGuardForwardsModeUntilLexicalScopeExit` to `test_a5_fatal.cpp`: `FakeRuntime` gains `pending_scope_mode` (immediately after `ops`, the prefix the inline wrappers pun through) and `last_scope_mode`; `fake_scope_begin` records the mode; the test asserts AUTO then MANUAL forwarding across lexical scope exit. Both test files also extend the single `offsetof(ops) == 0` guard to a second one pinning `pending_scope_mode` at `offsetof(PTO2Runtime, pending_scope_mode)`, directly asserting the fake matches the production `PTO2Runtime` prefix layout the inline wrappers pun through (not just pointer size). Both tests also seed `pending_scope_mode = PTO2ScopeMode::MANUAL` before the first guard so the `AUTO` assertion actually verifies the guard writes the default, rather than reading a pre-existing `AUTO` from the member initializer. R2 — a2a3 `RUNTIME_LOGIC.md` still described the retained temp buffer as a single runner-owned instance, but the shared host-runtime callback `get_retained_temp_buffer(runner_ctx, pipeline_slot, ...)` is per-slot and a5's matching paragraph already says so. Align a2a3 to a5's wording ("each slot's buffer lives on the runner"). Also fix a stale accessor location in both trees' `MULTI_RING.md`: `PTO2TaskId`'s direct accessors live in `src/common/task_interface/pto_task_id.h`, not `pto_runtime2_types.h`. Verification: built and ran `test_a2a3_fatal` and `test_a5_fatal` — 4/4 tests pass each (including the new a5 ScopeGuard case); both files are clang-format clean (ColumnLimit 120).
ChaoZheng109
pushed a commit
that referenced
this pull request
Aug 10, 2026
…#1752) Close the two P4/P5 residuals flagged in #1582's 2026-08-09 re-audit (after #1718 landed): R1 — a5 lacked the scope-guard regression test that a2a3 gained in #1718. a5's `PTO2_SCOPE_GUARD(...)` has been variadic (and its MANUAL path live, read by `rt_scope_begin` -> `begin_scope(mode)`) since the #1100 import, but `ut-a5` would not have caught a regression in MANUAL forwarding. Mirror `ScopeGuardForwardsModeUntilLexicalScopeExit` to `test_a5_fatal.cpp`: `FakeRuntime` gains `pending_scope_mode` (immediately after `ops`, the prefix the inline wrappers pun through) and `last_scope_mode`; `fake_scope_begin` records the mode; the test asserts AUTO then MANUAL forwarding across lexical scope exit. Both test files also extend the single `offsetof(ops) == 0` guard to a second one pinning `pending_scope_mode` at `offsetof(PTO2Runtime, pending_scope_mode)`, directly asserting the fake matches the production `PTO2Runtime` prefix layout the inline wrappers pun through (not just pointer size). Both tests also seed `pending_scope_mode = PTO2ScopeMode::MANUAL` before the first guard so the `AUTO` assertion actually verifies the guard writes the default, rather than reading a pre-existing `AUTO` from the member initializer. R2 — a2a3 `RUNTIME_LOGIC.md` still described the retained temp buffer as a single runner-owned instance, but the shared host-runtime callback `get_retained_temp_buffer(runner_ctx, pipeline_slot, ...)` is per-slot and a5's matching paragraph already says so. Align a2a3 to a5's wording ("each slot's buffer lives on the runner"). Also fix a stale accessor location in both trees' `MULTI_RING.md`: `PTO2TaskId`'s direct accessors live in `src/common/task_interface/pto_task_id.h`, not `pto_runtime2_types.h`. Verification: built and ran `test_a2a3_fatal` and `test_a5_fatal` — 4/4 tests pass each (including the new a5 ScopeGuard case); both files are clang-format clean (ColumnLimit 120).
This was referenced Aug 12, 2026
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
2 participants
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.
Summary
Converge the remaining low-risk A2/A3 and A5
tensormap_and_ringbufferruntime differences identified in #1582.What changed
PTO2_SCOPE_GUARD(...)to forwardPTO2ScopeMode, with AUTO/MANUAL lifetime coverage.PTO2TaskPayload::prefetch(), remove the duplicate helper, dead A2/A3 aliases/overloads, stale internal names, and A5's always-zerosched_scan_cycleleftover from the cleanup scoped to A2/A3 in Refactor: drop SCHED_IDLE_WAIT phase records and vestigial SCAN slot #869.#pragma onceand correct staleinit_data_from_layoutreferences.The cleanup is intentionally limited to the TMR runtime;
host_build_graphis unchanged.Verification
pre-commit run --files $(git diff --name-only upstream/main...)passed, including clang-tidy, cpplint, and markdownlint.ctest -LE requires_hardware --output-on-failure: 89/89 passed.git diff --checkand residual-symbol/header scans passed.mkdocs build --strictwas not available locally becausemkdocsis not installed in the current environment; Markdown lint passed and CI will run the repository docs job.Related to #1582.