[rocjitsu] Apply async IOFFSET to LDS addresses by kuhar · Pull Request #10199 · ROCm/rocm-systems

kuhar · 2026-08-15T11:42:21Z

Apply the signed 24-bit IOFFSET to both the global and LDS addresses for ordinary and cluster async loads and async stores, matching CDNA5 section 10.8.1 and expressions 95-102 and 106-109.

Reject LDS destinations outside the issuing workgroup allocation, keep cluster remapping widened through validation, and filter invalid lanes from writeback and race detection.

Add generator, execution, multicast, and plugin coverage for signed offsets, allocation bounds, and wraparound addresses.

Fixes #8829

therock-pr-bot · 2026-08-15T11:43:13Z

✅ All Checks Passed — Ready for Review

Check	Status	Details
📝 PR Description	✅ Pass	—
⛔ Forbidden Files	✅ Pass	—
🧪 Unit Test	✅ Pass	—
🔎 pre-commit	✅ Pass	—
🚫 Draft PR	🔜 To Be Enabled	—
🚩 Feature Flag	🔜 To Be Enabled	—
📊 Code Coverage	🔜 To Be Enabled	—
🤖 therock-pr-bot	✅ Pass	—

🎉 All checks passed! This PR is ready for review.

📖 Need help? See the Policy FAQ for details on every check and how to fix failures.

🙋 Wish to Override Policy?

therock-pr-bot · 2026-08-15T11:43:31Z

🎉 All checks passed! This PR is ready for review.

Copilot

Pull request overview

This PR updates rocJITsu’s CDNA5 async-to-LDS address semantics so the signed 24-bit IOFFSET is applied consistently to both global and LDS sides for async loads/stores (including cluster variants), while also hardening LDS bounds/overflow handling and ensuring invalid lanes are dropped from writeback and race detection.

Changes:

Apply signed 24-bit IOFFSET uniformly via a shared helper (signed_ioffset) and compute async LDS destinations with allocation-bounds validation (async_lds_lane_address).
Introduce and propagate an explicit per-workgroup LDS allocation size on Wavefront (lds_size) so address validation can reject out-of-range destinations.
Widen/validate cluster LDS multicast address computations and drop invalid/out-of-range per-lane targets; extend unit/integration/plugin/codegen tests for offsets, wraparound, and lane dropping.

Reviewed changes

Copilot reviewed 16 out of 17 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
emulation/rocjitsu/tests/execution_plugin_test.cpp	Adds plugin coverage ensuring dropped async LDS lanes do not create spurious low-address races.
emulation/rocjitsu/tests/cdna5_instruction_test.cpp	Ensures instruction tests initialize `Wavefront` LDS allocation size used by new address validation.
emulation/rocjitsu/tests/cdna5_dispatch_memory_test.cpp	Adds LDS wraparound-range rejection test exercising overflow-safe bounds checks.
emulation/rocjitsu/tests/cdna5_cluster_test.cpp	Expands decode/execute/simulation tests for signed `IOFFSET`, bounds rejection, and cluster multicast dropping/overflow handling.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/plugins/race_detector/plugin.cpp	Filters invalid LDS lane addresses from race detector lane masks.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/amdgpu/wavefront.h	Adds `lds_size` tracking on wavefronts for per-workgroup LDS allocation bounds.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/amdgpu/spi.h	Initializes `Wavefront::lds_size` during dispatch.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/amdgpu/memory_pipeline.cpp	Skips writeback and improves trace output for invalid per-lane LDS destinations.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/amdgpu/lds.h	Introduces `kInvalidLdsAddress` sentinel and overflow-safe bounds checking helpers.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/amdgpu/command_processor.cpp	Initializes `Wavefront::lds_size` for CP-dispatched workgroups.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/amdgpu/cluster_lds_multicast.h	Widens remap/lane-address helpers to `uint64_t` to detect/handle overflow.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/vm/amdgpu/cluster_lds_multicast.cpp	Drops invalid/out-of-range per-lane multicast targets and validates widened addresses.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/isa/arch/amdgpu/generated/cdna5/vglobal_exec.cpp	Switches to `signed_ioffset` and uses `async_lds_lane_address` for async LDS operands.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/isa/arch/amdgpu/cdna5/addr_calc.h	Adds `signed_ioffset` and declares `async_lds_lane_address`.
emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/isa/arch/amdgpu/cdna5/addr_calc.cpp	Implements allocation-bounded async LDS address computation; reuses `signed_ioffset` broadly.
emulation/rocjitsu/lib/python/amdisa/tests/test_generator_profile_gates.py	Adds generator test assertions ensuring new helpers are emitted in generated bodies.
emulation/rocjitsu/lib/python/amdisa/codegen/_generator.py	Updates codegen to emit `async_lds_lane_address` and `signed_ioffset` usage and revised comments.

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

Apply the signed 24-bit IOFFSET to both the global and LDS addresses for ordinary and cluster async loads and async stores, matching CDNA5 section 10.8.1 and expressions 95-102 and 106-109. Reject LDS destinations outside the issuing workgroup allocation, keep cluster remapping widened through validation, and filter invalid lanes from writeback and race detection. Add generator, execution, multicast, and plugin coverage for signed offsets, allocation bounds, and wraparound addresses. Fixes #8829

atgutier

LGTM. Just one comment that could harden our bounds check.

Apply the signed 24-bit IOFFSET to both the global and LDS addresses for ordinary and cluster async loads and async stores, matching CDNA5 section 10.8.1 and expressions 95-102 and 106-109. Reject LDS destinations outside the issuing workgroup allocation, keep cluster remapping widened through validation, and filter invalid lanes from writeback and race detection. Add generator, execution, multicast, and plugin coverage for signed offsets, allocation bounds, and wraparound addresses. Fixes #8829 (cherry picked from commit be1f919)

…t calls (#10444) ## Motivation hipTensor and rocThrust gfx1250 objects are refused at code-object load time with "indirect branch or call target recovery is not implemented for relocated kernel text" on `s_swap_pc_i64`. A refused object leaves HIP holding an invalid kernel handle, so the failure surfaces as a segfault rather than a diagnostic. The fix is upstream as #10376; this backports it to `release/bkc/therock-10.1-20260811`. ## Commits Cherry-Picked | # | Upstream | PR | Backport | Subject | |---|----------|-----|----------|---------| | 1 | `b6d838028e` | #10197 | `846fb16d12` | [rocjitsu][docs] Refresh architecture support documentation | | 2 | `5c8e33e391` | #10200 | `db5c58b1be` | [rocjitsu] Define standalone decoder ownership | | 3 | `be1f91998f` | #10199 | `ed102baf1c` | [rocjitsu] Apply async IOFFSET to LDS addresses | | 4 | `fda635e5b4` | — | `ad32ff7c18` | [rocjitsu] Extract per-scope rollback and object-level checks from translate() | | 5 | `8f75eb781e` | — | `4c0cb05207` | [rocjitsu] Emit leading V_NOPs ahead of a gfx1250 MODE write | | 6 | `00aa1e260f` | #9941 | `ded43b1787` | fix(rocjitsu): disable zero-count TDM descriptors | | 7 | `c810647609` | #10192 | `50ca7f1732` | [rocjitsu] Apply MUBUF LDS-load instruction offset to the LDS address | | 8 | `1d7866269d` | #9560 | `2e6bf26943` | [rocjitsu] code rewrite condition validator | | 9 | `95b6c410bb` | #10175 | `8418901076` | [rocjitsu] Fix ds_read_b64_tr_b16 lane layout in gfx950 simulator | | 10 | `bfaebcfb5c` | #10240 | `c4eac1e1f7` | [rocjitsu][util] Add result and diagnostic utilities | | 11 | `dbb7221fa7` | #9775 | `cdd0993a5f` | fix(rocjitsu): route tensor DMA through wave VMID | | 12 | `2779247222` | #9779 | `7bfeba6f5d` | feat(rocjitsu): Lazy vgpr allocation system | | 13 | `a659172a31` | #9936 | `180cf00325` | fix(rocjitsu): honor TDM tensor bounds across iterations | | 14 | `60c29efb39` | #10204 | `370bfb9796` | [rocjitsu] Fix gfx950 decoder extension handling | | 15 | `c1dba564e7` | #10345 | `536110a083` | [rocjitsu] Model gfx1250 wave ABI state | | 16 | `7f71be9f32` | #10346 | `3f922ad5f2` | [rocjitsu] Model gfx1250 barrier synchronization | | 17 | `bea7ac8937` | #10368 | `7b5ecf6c7b` | [rocjitsu] Replace decoder exceptions with results | | 18 | `f2e878694e` | #10371 | `8cf750d23f` | [rocjitsu] Dispatch operand validation by type | | 19 | `dd619bed77` | #9735 | `1a612c47c0` | fix(rocjitsu): respect same-wave LDS ordering | | 20 | `c5fae8d2b2` | #10137 | `4360c4b978` | [rocjitsu] Rename gfx1250 to cdna5 for enums, configs, amdisa codegen helpers and generated symbols. (NFC) | | 21 | `b05a587f89` | #10274 | `bfb5f9fab0` | [rocjitsu] Align CDNA4 MI350X configs with captured topology | | 22 | `f81f9d40db` | #10362 | `8177b5725c` | feat(rocjitsu): cdna5 OOB behavior implementation | | 23 | `53a939bd00` | #10283 | `141b109432` | [rocjitsu] add per-XCD workgroup dispatch observability | | 24 | `a0fb11bc17` | #10414 | `6bb4c5a959` | [rocjitsu] Refactor test encoding handling in execution tests (NFC) | | 25 | `e5109e315e` | #9844 | `322ac26b44` | feat(emulation): debug emulated GPU kernels with real ROCgdb | | 26 | `efa24f23c9` | — | `c3c9641073` | rocjitsu: Point the XCD distribution test at the config that exists | | 27 | `8b7763ef5e` | #10376 | `c3f419555a` | [rocjitsu] Translate gfx1250 lane-banked and kernarg-supplied indirect calls | ## Issue Tracking [ROCM-29436] ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/rocm-systems/blob/develop/CONTRIBUTING.md. [ROCM-29346]: https://amd-hub.atlassian.net/browse/ROCM-29346?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ [ROCM-29436]: https://amd-hub.atlassian.net/browse/ROCM-29436?atlOrigin=eyJpIjoiNWRkNTljNzYxNjVmNDY3MDlhMDU5Y2ZhYzA5YTRkZjUiLCJwIjoiZ2l0aHViLWNvbS1KU1cifQ --------- Signed-off-by: Tony Gutierrez <anthony.gutierrez@amd.com> Signed-off-by: hanhanW <hanhan0912@gmail.com> Co-authored-by: Jakub Kuderski <jakub@nod-labs.com> Co-authored-by: James Newling <james.newling@gmail.com> Co-authored-by: Fei Kou <feikou@meta.com> Co-authored-by: Austin Lu <austin.lu@amd.com> Co-authored-by: Fei Kou <feikou@anthropic.com> Co-authored-by: Erick Ochoa Lopez <eochoalo@amd.com> Co-authored-by: Han-Chung Wang <hanhan0912@gmail.com> Co-authored-by: Rosa, Ada <arosa@amd.com> Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

Copilot AI lite review requested due to automatic review settings August 15, 2026 11:42

kuhar requested review from a team and atgutier as code owners August 15, 2026 11:42

github-actions Bot added the project: rocjitsu label Aug 15, 2026

Copilot started reviewing on behalf of kuhar August 15, 2026 11:43

View session

Copilot AI reviewed Aug 15, 2026

View reviewed changes

kuhar force-pushed the users/kuhar/gfx1250-cluster-lds-offset branch from 4b77b8d to 0a9b79a Compare August 15, 2026 11:50

systems-assistant Bot added the organization: ROCm label Aug 15, 2026

atgutier approved these changes Aug 15, 2026

View reviewed changes

Comment thread emulation/rocjitsu/lib/rocjitsu/src/rocjitsu/isa/arch/amdgpu/cdna5/addr_calc.cpp Resolved

kuhar force-pushed the users/kuhar/gfx1250-cluster-lds-offset branch from 0a9b79a to becb955 Compare August 16, 2026 00:05

kuhar merged commit be1f919 into develop Aug 16, 2026
27 checks passed

kuhar deleted the users/kuhar/gfx1250-cluster-lds-offset branch August 16, 2026 00:46

atgutier mentioned this pull request Aug 19, 2026

[rocjitsu] Translate gfx1250 lane-banked and kernarg-supplied indirect calls #10444

Merged

1 task done

peymanr mentioned this pull request Nov 6, 2026

[rocjitsu] Add CDNA3/CDNA4 global_load_lds_* execution support #14867

Open

Navigation Menu

Repository navigation

[rocjitsu] Apply async IOFFSET to LDS addresses - #10199

[rocjitsu] Apply async IOFFSET to LDS addresses#10199
kuhar merged 1 commit into
developfrom
users/kuhar/gfx1250-cluster-lds-offset

kuhar commented Aug 15, 2026

Uh oh!

therock-pr-bot Bot commented Aug 15, 2026 •

edited

Loading

Uh oh!

therock-pr-bot Bot commented Aug 15, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

atgutier left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

kuhar commented Aug 15, 2026

Uh oh!

therock-pr-bot Bot commented Aug 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ All Checks Passed — Ready for Review

Uh oh!

therock-pr-bot Bot commented Aug 15, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

atgutier left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

therock-pr-bot Bot commented Aug 15, 2026 •

edited

Loading