CMA-memory-to-offload-FNN-PLUS-module-to-eNPU by qti-sbojja · Pull Request #2159 · qualcomm-linux/kernel-topics

qti-sbojja · 2026-10-29T11:20:04Z

This series contains patches for allocating CMA memory to offload
FNN PLUS module to eNPU.

CRs-Fixed: 4727557
JIRA: https://jira-dc.qualcomm.com/jira/browse/QLIJIRA-210

qlijarvis · 2026-10-29T11:40:33Z

🔨 Build Failure Analysis — PR #2159

PR: #2159
Build run: https://github.com/qualcomm-linux/kernel-config/actions/runs/40865205339

#	Error	File:Line	PR-introduced?	Root Cause
1	DTS syntax error	`arch/arm64/boot/dts/qcom/qcs6490-rb3gen2.dts:393`	Yes	`audio_cma_mem` node has incorrect indentation/placement in device tree hierarchy
2	Phandle reference error	`arch/arm64/boot/dts/qcom/qcs6490-audioreach.dtsi:97`	Yes	Reference to `audio_cma_mem` fails because node definition has syntax error

Verdict

Both errors are introduced by this PR. The root cause is incorrect device tree node placement in patch 1/2 - the audio_cma_mem node is not properly nested within a reserved-memory parent node.

📎 Detailed analysis: Full report

qlijarvis · 2026-10-29T11:45:36Z

🔨 Build Failure Analysis — PR #2159

PR: #2159
Build run: https://github.com/qualcomm-linux/kernel-config/actions/runs/40865205339

#	Error	File:Line	PR-introduced?	Root Cause
1	DTS syntax error	`qcs6490-rb3gen2.dts:393`	Yes	Incorrect indentation/placement of `audio_cma_mem` node - appears to be nested inside wrong parent node instead of being a sibling in `reserved-memory`
2	Phandle reference error	`qcs6490-audioreach.dtsi:97`	Yes	Reference to `audio_cma_mem` label fails because the node definition has a syntax error and cannot be parsed

Verdict

Both errors are introduced by this PR due to incorrect device tree node placement in patch 1/2.

📎 Detailed analysis: Full report

qlijarvis · 2026-10-29T12:08:47Z

PR #2159 — validate-patch

PR: #2159

Verdict	Issues	Detailed Report
❌	0	Full report

Final Summary

Lore link present: Yes — https://lore.kernel.org/all/20260825133333.376591-1-karthik.s@oss.qualcomm.com/
Lore link matches PR commits: No — Patch 1 has critical placement error (node inside wrong parent) and deprecated node naming. Patch 2 has different base tree but semantically identical diff content.
Upstream patch status: ⏳ Decision Pending (NACKed) — Maintainers rejected patch 1 for deprecated node naming and lack of compile testing. Series needs v2 revision.
PR present in qcom-next/topics: Fail - 2/2 commit(s) are missing from both qcom-next and topics

Verdict: ❌ — click to expand

🔍 Patch Validation

PR: #2159 — CMA memory to offload FNN PLUS module to eNPU
Upstream commit: https://lore.kernel.org/all/20260825133333.376591-1-karthik.s@oss.qualcomm.com/
Verdict: ❌ FAIL

Commit Message

Check Status Note

Subject matches upstream ✅ Both commits use FROMLIST: prefix correctly

Body preserves rationale ✅ Technical rationale preserved in both commits

Fixes tag present/correct N/A No Fixes tag in upstream (new feature)

Authorship preserved ✅ From: matches lore author (Karthik S)

Backport note (if applicable) N/A FROMLIST: — not yet merged upstream

Link tag present ✅ Both commits have correct lore Link

Diff

Commit 1/2: qcs6490-rb3gen2 audio CMA

File Status Notes

qcs6490-audioreach.dtsi ✅ Identical — adds memory-region = <&audio_cma_mem>;

qcs6490-rb3gen2.dts ❌ Context mismatch — PR has different base tree

Lore patch 1:
@@ -174,6 +174,14 @@ rmtfs_mem: memory@f8500000 {
 			qcom,client-id = <1>;
 			qcom,vmid = <QCOM_SCM_VMID_MSS_MSA>, <QCOM_SCM_VMID_NAV>;
 		};
+
+		audio_cma_mem: qcom,audio-ml {
PR patch 1:
@@ -172,6 +172,13 @@
 
 			qcom,client-id = <1>;
 			qcom,vmid = <QCOM_SCM_VMID_MSS_MSA>, <QCOM_SCM_VMID_NAV>;
+
+		audio_cma_mem: qcom,audio-ml {
The PR patch applies at line 172 (missing closing brace for rmtfs_mem node), while lore applies at line 174 (after closing brace). The PR patch has incorrect indentation and placement — the audio_cma_mem node is placed inside the rmtfs_mem node instead of as a sibling in the reserved-memory parent.

Commit 2/2: kodiak-el2 IOMMU SIDs

File Status Notes

kodiak-el2.dtso ❌ Index hash mismatch — different base tree

Lore patch 2:
index 8daf7f02fd75..d050857cab31 100644
PR patch 2:
index 05e1b696a6206..8851588e3e83d 100644
The diff content (adding three IOMMU SIDs) is semantically identical, but the base tree differs.

Upstream Patch Status

Community verdict: ⏳ Decision Pending (NACKed with actionable feedback)

Krzysztof Kozlowski (DT maintainer) replied on 2026-08-25:

On patch 1/2:
"Don't send downstream code. There are no such patterns in upstream for years."
(Referring to the node name qcom,audio-ml — vendor prefix in node name is deprecated)

Follow-up:
"And the 'equivalent implementation for lemans' received comments which you missed and re-implemented the same issues."

Konrad Dybcio (Qualcomm maintainer) replied:

"Please don't send patches that you haven't compile-tested"

Analysis:

Patch 1 has a node naming issue (qcom,audio-ml should be audio-ml or similar without vendor prefix in the node name part after the label).

Maintainers flagged that this repeats issues from a previous lemans (SA8295P) series.

Konrad's comment suggests the patch does not compile (likely due to the indentation/placement error visible in the PR diff).

Recommendation: The series needs revision (v2) addressing:

Remove vendor prefix from node name (audio_cma_mem: audio-ml not qcom,audio-ml)

Fix the indentation/placement of the audio_cma_mem node in patch 1

Verify the patch compiles cleanly

Issues

Patch 1 — Incorrect node placement (CRITICAL):
The PR patch places audio_cma_mem inside the rmtfs_mem node (line 172, missing closing brace context) instead of as a sibling in reserved-memory. The lore patch correctly places it after the rmtfs_mem closing brace (line 174).
Impact: This will cause a DT syntax error or incorrect memory reservation hierarchy.

Patch 1 — Deprecated node naming (upstream blocker):
Node name qcom,audio-ml uses vendor prefix, which maintainers explicitly rejected. Should be audio-ml or audio-mem-cma.

Base tree mismatch:
Both patches apply to a different base than the lore submission (different index hashes). This is expected for a vendor tree but means the PR is not a faithful cherry-pick of the lore posting.

Compile failure (per maintainer feedback):
Konrad Dybcio stated the patch was not compile-tested, suggesting it has build errors (consistent with the indentation issue found).

Verdict

❌ FAIL — Do not merge without correction.

Required fixes:

Patch 1: Correct the indentation and placement of audio_cma_mem node — it must be a sibling of rmtfs_mem in the reserved-memory parent, not a child of rmtfs_mem.

Patch 1: Change node name from qcom,audio-ml to audio-ml (or audio-mem-cma) to comply with upstream DT naming conventions.

Both patches: Rebase to the correct base tree (or document the base tree difference if intentional for vendor integration).

Both patches: Compile-test before resubmission.

Final Summary

Lore link present: Yes — https://lore.kernel.org/all/20260825133333.376591-1-karthik.s@oss.qualcomm.com/

Lore link matches PR commits: No — Patch 1 has critical placement error (node inside wrong parent) and deprecated node naming. Patch 2 has different base tree but semantically identical diff content.

Upstream patch status: ⏳ Decision Pending (NACKed) — Maintainers rejected patch 1 for deprecated node naming and lack of compile testing. Series needs v2 revision.

PR present in qcom-next/topics: No — Per integration_presence_report.md, both commits are missing from qcom-next and all topic branches (0/2 present). This is expected since the upstream series has not been accepted yet.

Deterministic Integration Presence

Integration Presence Report

This report is generated by Jarvis before validate-patch runs.
It is the authoritative source for whether PR changes are already present
in qcom-next or in the kernel topic branches.

Kernel repo: /local/mnt/workspace/sgaud/Qgenie/image_pipeline/kernel
qcom-next ref: c7896917e523946138ce654d4b6c1e33d5a8f4e5
topics remote: topics -> https://github.com/qualcomm-linux/kernel-topics
topics fetch: fetched

Commit Subject qcom-next topics Final

1/2 [PATCH 1/2] FROMLIST: arm64: dts: qcom: qcs6490-rb3gen2: add audio missing - no subject, patch-id, or full tree-content match found missing - no subject, patch-id, or full tree-content match found missing

2/2 [PATCH 2/2] FROMLIST: arm64: dts: qcom: kodiak-el2: add LPASS AML missing - no subject, patch-id, or full tree-content match found missing - no subject, patch-id, or full tree-content match found missing

Final Status

overall_status: FAIL
present_commits: 0/2
partial_commits: 0/2
missing_commits: 2/2
topics_checked_for_commits: 2/2
final_summary: PR present in qcom-next/topics: Fail - 2/2 commit(s) are missing from both qcom-next and topics

qlijarvis · 2026-10-29T12:08:49Z

PR #2159 — checker-log-analyzer

PR: #2159
Checker run: https://github.com/qualcomm-linux/kernel-config/actions/runs/40865204435

Checker	Result	Summary
Checker	Result	Summary
checkpatch	⏭️	Skipped - automerge failed
dt-binding-check	⏭️	Skipped - automerge failed
dtb-check	⏭️	Skipped - automerge failed
sparse-check	⏭️	Skipped - automerge failed
check-uapi-headers	⏭️	Skipped - automerge failed
check-patch-compliance	⏭️	Skipped - automerge failed
tag-check	✅	Both commits have valid `FROMLIST:` prefix
qcom-next-check	N/A	Target branch is qcom-next (topic branch)

Detailed report: Full report

Checker analysis — click to expand

🤖 CI Checker Analysis (checker-log-analyzer)

PR: #2159 - FROMLIST: arm64: dts: qcom: qcs6490-rb3gen2 audio CMA + kodiak-el2 LPASS IOMMU
Source: https://github.com/qualcomm-linux/kernel-config/actions/runs/40865204435

Checker Result Summary

checkpatch ⏭️ Skipped - automerge failed

dt-binding-check ⏭️ Skipped - automerge failed

dtb-check ⏭️ Skipped - automerge failed

sparse-check ⏭️ Skipped - automerge failed

check-uapi-headers ⏭️ Skipped - automerge failed

check-patch-compliance ⏭️ Skipped - automerge failed

tag-check ✅ Both commits have valid FROMLIST: prefix

qcom-next-check N/A Target branch is qcom-next (topic branch)

❌ Automerge Failure (Pre-Checker Stage)

Root cause: Merge conflicts during CI automerge step prevented all checker jobs from running.

Failure details:
CONFLICT (content): Merge conflict in arch/arm64/boot/dts/qcom/Makefile
CONFLICT (add/add): Merge conflict in arch/arm64/boot/dts/qcom/kodiak-el2.dtso
Automatic merge failed; fix conflicts and then commit the result.
Merge failed, manual merge
Analysis:

The CI workflow attempts to merge the PR topic branch (topic/tech/all/dt/qcs6490) into the baseline integration branch (qcom-next at c7896917e523946138ce654d4b6c1e33d5a8f4e5) before running checkers. This merge failed with two conflicts:

arch/arm64/boot/dts/qcom/Makefile — Content conflict (likely overlapping dtb/dtso entries)

arch/arm64/boot/dts/qcom/kodiak-el2.dtso — Add/add conflict (both sides added the same file with different content)

The second conflict is particularly significant: the PR modifies kodiak-el2.dtso to add LPASS AML IOMMU SIDs, but the baseline integration branch already contains a version of this file with different content. This is a classic "both modified" or "both added" scenario where the file exists in both branches with incompatible changes.

Why this happened:

The PR is based on an older snapshot of the topic branch

The baseline integration branch (qcom-next) has evolved since the PR was created

Another change to kodiak-el2.dtso or Makefile was merged into the baseline after this PR's base commit

Fix:

The PR author must rebase the topic branch onto the current qcom-next tip and resolve the conflicts manually:
# In the kernel-topics repo
git fetch origin qcom-next
git checkout topic/tech/all/dt/qcs6490
git rebase origin/qcom-next

# Resolve conflicts in:
#   arch/arm64/boot/dts/qcom/Makefile
#   arch/arm64/boot/dts/qcom/kodiak-el2.dtso

# For Makefile: merge the dtb/dtso entries from both sides
# For kodiak-el2.dtso: integrate the IOMMU SID changes with the baseline version

git add arch/arm64/boot/dts/qcom/Makefile arch/arm64/boot/dts/qcom/kodiak-el2.dtso
git rebase --continue

# Force-push the rebased topic branch
git push --force-with-lease origin topic/tech/all/dt/qcs6490
Reproduce locally:
git clone https://github.com/qualcomm-linux/kernel.git
cd kernel
git fetch origin qcom-next
git checkout -b test-merge origin/qcom-next
git fetch https://github.com/qualcomm-linux/kernel-topics.git topic/tech/all/dt/qcs6490
git merge FETCH_HEAD
# Conflicts will appear — inspect with:
git diff --name-only --diff-filter=U
✅ tag-check (Manual Verification)

Status: PASS

Both commits in the PR have valid subject-line prefixes:

FROMLIST: arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA memory region for AudioReach

FROMLIST: arm64: dts: qcom: kodiak-el2: add LPASS AML IOMMU SIDs for eNPU audio offload

Both commits include Link: trailers pointing to the upstream lore.kernel.org submission:
Link: https://lore.kernel.org/all/20260825133333.376591-1-karthik.s@oss.qualcomm.com/
Note: The target branch is qcom-next (via topic branch topic/tech/all/dt/qcs6490), so the subject-prefix check is not mandatory per the skill rules. However, the commits correctly use FROMLIST: which is best practice for patches posted upstream.

📋 Patch Content Summary

Commit 1/2: Add audio CMA memory region for AudioReach on qcs6490-rb3gen2

Adds a 16 MiB dynamic CMA pool with 4 MiB alignment

Uses alloc-ranges to avoid EFI-reserved memory

Wires memory-region into q6apmdai for the q6apm-audio-mem-cma driver

Commit 2/2: Add LPASS AML IOMMU SIDs for eNPU audio offload on kodiak-el2

Adds three LPASS AML stream IDs to remoteproc_adsp

Enables IOMMU domain for CMA device (fixes has_iommu=0 issue)

Required for IOCTL_MAP_HYP_ASSIGN when Gunyah owns the SMMU

Verdict

Cannot merge — automerge conflicts must be resolved before checkers can run.

Action required:

Rebase topic/tech/all/dt/qcs6490 onto current qcom-next tip

Resolve conflicts in Makefile and kodiak-el2.dtso

Force-push the rebased topic branch

Re-trigger CI to run checkers on the conflict-free merge

Once conflicts are resolved: All checkers should run cleanly (DTS-only changes typically pass checkpatch, dt-binding-check, and dtb-check with no issues, and sparse/uapi-headers will skip).

qlijarvis · 2026-10-30T05:49:24Z

🔨 Build Failure Analysis — PR #2159

PR: #2159
Build run: https://github.com/qualcomm-linux/kernel-config/actions/runs/41008485358

#	Error	File:Line	PR-introduced?	Root Cause
1	ERROR (phandle_references): Reference to non-existent node or label "audio_cma_mem"	arch/arm64/boot/dts/qcom/qcs6490-audioreach.dtsi:97	Yes	Patch adds `memory-region = <&audio_cma_mem>` to shared .dtsi file, but only defines `audio_cma_mem` node in qcs6490-rb3gen2.dts. Other boards (qcm6490-idp, qcs6490-radxa-dragon-q6a) that include this .dtsi lack the node definition.

Verdict

1 of 1 error is introduced by this PR. The error affects both standard and RT kernel builds, causing DTB compilation to fail for qcm6490-idp.dtb and qcs6490-radxa-dragon-q6a.dtb.

📎 Detailed analysis: Full report

qlijarvis · 2026-10-30T05:54:17Z

🔨 Build Failure Analysis — PR #2159

PR: #2159
Build run: https://github.com/qualcomm-linux/kernel-config/actions/runs/41008485358

#	Error	File:Line	PR-introduced?	Root Cause
1	ERROR (phandle_references): Reference to non-existent node or label "audio_cma_mem"	arch/arm64/boot/dts/qcom/qcs6490-audioreach.dtsi:100	Yes	Patch adds `memory-region = <&audio_cma_mem>` to shared include file but only defines `audio_cma_mem` in qcs6490-rb3gen2.dts, breaking qcm6490-idp.dtb and qcs6490-radxa-dragon-q6a.dtb builds

Verdict

1 devicetree compilation error is introduced by this PR (affects 2 board DTB files in both standard and RT builds).

📎 Detailed analysis: Full report

qlijarvis · 2026-10-30T06:26:22Z

PR #2159 — validate-patch

PR: #2159

Verdict	Issues	Detailed Report
❌	2	Full report

Final Summary

Lore link present: Yes — https://lore.kernel.org/all/20260825133333.376591-1-karthik.s@oss.qualcomm.com/
Lore link matches PR commits: Partial — Patch 2/2 matches exactly; Patch 1/2 matches the rejected v1 content (node name qcom,audio-ml violates DT conventions per maintainer feedback)
Upstream patch status: ❌ NACKed — Patch 1/2 explicitly rejected by Krzysztof Kozlowski for downstream node naming pattern; Patch 2/2 pending but blocked by 1/2 rejection
PR present in qcom-next/topics: Fail - 2/2 commit(s) are missing from both qcom-next and topics

Verdict: ❌ — click to expand

🔍 Patch Validation

PR: #2159 - "CMA-memory-to-offload-FNN-PLUS-module-to-eNPU"
Upstream commit: https://lore.kernel.org/all/20260825133333.376591-1-karthik.s@oss.qualcomm.com/
Verdict: ❌ FAIL

Commit Message

Check Status Note

Subject matches upstream ✅ Both commits match lore subjects exactly

Body preserves rationale ✅ Technical rationale preserved in both commits

Fixes tag present/correct N/A No Fixes tag in upstream or PR (new feature)

Authorship preserved ✅ From: Karthik S <karthik.s@oss.qualcomm.com> matches lore author

Backport note (if applicable) N/A FROMLIST: prefix used correctly - not a backport

Diff

File Status Notes

arch/arm64/boot/dts/qcom/qcs6490-audioreach.dtsi ✅ Identical to lore patch 1/2

arch/arm64/boot/dts/qcom/qcs6490-rb3gen2.dts ❌ Node name violates upstream DT naming convention

arch/arm64/boot/dts/qcom/kodiak-el2.dtso ✅ Identical to lore patch 2/2

Issues

Commit 1/2 — Critical upstream rejection:
Node name qcom,audio-ml violates devicetree naming conventions
Lore patch uses: audio_cma_mem: qcom,audio-ml {

Krzysztof Kozlowski (DT maintainer) explicitly rejected this pattern on Aug 25, 2026:

"Don't send downstream code. There are no such patterns in upstream for years."
[Source: lore thread reply]

Root cause: Node names must not contain vendor prefixes. The qcom, prefix belongs in the compatible property, not the node name.
Correct pattern: Node name should be generic and describe the hardware function:
audio_cma_mem: linux,cma {
    compatible = "shared-dma-pool";
    ...
};
or simply:
audio_cma_mem: memory@0 {
    compatible = "shared-dma-pool";
    ...
};
Maintainer also noted missing compile-testing

Konrad Dybcio replied: "Please don't send patches that you haven't compile-tested"

This suggests the lore patch may have DT schema validation errors
Commit 2/2:

Diff content is faithful to lore patch 2/2

No issues detected in diff comparison

Upstream Patch Status

Commit Community Verdict

[PATCH v1 1/2] arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA memory region ❌ NACKed — Krzysztof Kozlowski on Aug 25, 2026: "Don't send downstream code. There are no such patterns in upstream for years." Node name qcom,audio-ml violates DT naming conventions.

[PATCH v1 2/2] arm64: dts: qcom: kodiak-el2: add LPASS AML IOMMU SIDs ⏳ Decision Pending — No maintainer decision yet; last activity Aug 25, 2026. Depends on patch 1/2 being fixed first.

Evidence from lore thread:

Cover letter notes dependency: "This patch depends on: arm64: dts: qcom: kodiak: EL2 overlay for qcm6490-idp / qcs6490-rb3gen2"

Patch 1/2 received explicit rejection from DT maintainer

Patch 2/2 has no review comments yet, but is blocked by patch 1/2 rejection

Dependency Check

❌ Missing prerequisite: Cover letter states:

"This patch depends on: arm64: dts: qcom: kodiak: EL2 overlay for qcm6490-idp / qcs6490-rb3gen2
Link: https://lore.kernel.org/all/20260803121042.3313824-1-mukesh.ojha@oss.qualcomm.com/"

The PR does not include this prerequisite patch, and it's unclear if it's already merged in the target tree.

qcom-next/topics Presence

Commit Status

[PATCH 1/2] FROMLIST: arm64: dts: qcom: qcs6490-rb3gen2: add audio ⚠️ Not in qcom-next or topics

[PATCH 2/2] FROMLIST: arm64: dts: qcom: kodiak-el2: add LPASS AML ⚠️ Not in qcom-next or topics

Source: integration_presence_report.md — both commits missing from qcom-next (c7896917e523) and all topic branches.

Verdict

❌ DO NOT MERGE — This PR must be rejected and reworked.

Critical issues:

Upstream rejection: Patch 1/2 was explicitly NACKed by the devicetree maintainer for using downstream node naming patterns (qcom,audio-ml) that violate upstream conventions.

Incorrect FROMLIST: prefix: The PR uses FROMLIST: prefix, which implies the patch is under review and expected to be accepted upstream. However, patch 1/2 has been rejected and requires a v2 revision with corrected node naming.

Missing from integration branches: Neither commit is present in qcom-next or topic branches, confirming this is new code that hasn't been integrated yet.

Required actions:

Wait for upstream v2: The author must post a v2 series to lore.kernel.org with the node name fixed (remove qcom, prefix from node name, use generic name like linux,cma or memory@0).

Update PR after v2 acceptance: Once the v2 series receives maintainer approval (Reviewed-by/Acked-by tags), update this PR with the corrected patches and update the Link: tag to point to the v2 message-ID.

Verify prerequisite: Confirm that the kodiak EL2 overlay dependency (https://lore.kernel.org/all/20260803121042.3313824-1-mukesh.ojha@oss.qualcomm.com/) is either already merged or included in the PR.

Final Summary

Lore link present: Yes — https://lore.kernel.org/all/20260825133333.376591-1-karthik.s@oss.qualcomm.com/

Lore link matches PR commits: Partial — Patch 2/2 matches exactly; Patch 1/2 matches the rejected v1 content (node name qcom,audio-ml violates DT conventions per maintainer feedback)

Upstream patch status: ❌ NACKed — Patch 1/2 explicitly rejected by Krzysztof Kozlowski for downstream node naming pattern; Patch 2/2 pending but blocked by 1/2 rejection

PR present in qcom-next/topics: No — Both commits missing from qcom-next (c7896917e523) and all topic branches per integration_presence_report.md

Deterministic Integration Presence

Integration Presence Report

This report is generated by Jarvis before validate-patch runs.
It is the authoritative source for whether PR changes are already present
in qcom-next or in the kernel topic branches.

Kernel repo: /local/mnt/workspace/sgaud/Qgenie/image_pipeline/kernel
qcom-next ref: c7896917e523946138ce654d4b6c1e33d5a8f4e5
topics remote: topics -> https://github.com/qualcomm-linux/kernel-topics
topics fetch: fetched

Commit Subject qcom-next topics Final

1/2 [PATCH 1/2] FROMLIST: arm64: dts: qcom: qcs6490-rb3gen2: add audio missing - no subject, patch-id, or full tree-content match found missing - no subject, patch-id, or full tree-content match found missing

2/2 [PATCH 2/2] FROMLIST: arm64: dts: qcom: kodiak-el2: add LPASS AML missing - no subject, patch-id, or full tree-content match found missing - no subject, patch-id, or full tree-content match found missing

Final Status

overall_status: FAIL
present_commits: 0/2
partial_commits: 0/2
missing_commits: 2/2
topics_checked_for_commits: 2/2
final_summary: PR present in qcom-next/topics: Fail - 2/2 commit(s) are missing from both qcom-next and topics

qlijarvis · 2026-10-30T06:26:24Z

PR #2159 — checker-log-analyzer

PR: #2159
Checker run: https://github.com/qualcomm-linux/kernel-config/actions/runs/41008485321

Checker	Result	Summary
Checker	Result	Summary
checkpatch	⏭️	Skipped - automerge failed
dt-binding-check	⏭️	Skipped - automerge failed
dtb-check	⏭️	Skipped - automerge failed
sparse-check	⏭️	Skipped - automerge failed
check-uapi-headers	⏭️	Skipped - automerge failed
check-patch-compliance	⏭️	Skipped - automerge failed
tag-check	✅	Both commits have valid `FROMLIST:` prefix

Detailed report: Full report

Checker analysis — click to expand

🤖 CI Checker Analysis (checker-log-analyzer)

PR: #2159 - FROMLIST: arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA memory region for AudioReach
Source: https://github.com/qualcomm-linux/kernel-config/actions/runs/41008485321

Checker Result Summary

checkpatch ⏭️ Skipped - automerge failed

dt-binding-check ⏭️ Skipped - automerge failed

dtb-check ⏭️ Skipped - automerge failed

sparse-check ⏭️ Skipped - automerge failed

check-uapi-headers ⏭️ Skipped - automerge failed

check-patch-compliance ⏭️ Skipped - automerge failed

tag-check ✅ Both commits have valid FROMLIST: prefix

⚠️ Automerge Failure (Pre-Checker Stage)

Root cause: Merge conflicts during CI integration step prevented all checkers from running.

Failure details:
Merging topic branch: topic/tech/all/dt/qcs6490
Auto-merging arch/arm64/boot/dts/qcom/Makefile
CONFLICT (content): Merge conflict in arch/arm64/boot/dts/qcom/Makefile
Auto-merging arch/arm64/boot/dts/qcom/kodiak-el2.dtso
CONFLICT (add/add): Merge conflict in arch/arm64/boot/dts/qcom/kodiak-el2.dtso
Automatic merge failed; fix conflicts and then commit the result.
Analysis:

The CI workflow attempts to merge the PR branch (topic/tech/all/dt/qcs6490) into an integration branch based on baseline tag c7896917e523946138ce654d4b6c1e33d5a8f4e5. Two conflicts occurred:

arch/arm64/boot/dts/qcom/Makefile — Content conflict (likely overlapping DTB/DTSO additions)

arch/arm64/boot/dts/qcom/kodiak-el2.dtso — Add/add conflict (both branches added/modified this file independently)

The second conflict is particularly relevant because this PR modifies kodiak-el2.dtso (adding IOMMU SIDs to remoteproc_adsp). The "add/add" conflict indicates that another topic branch in the integration also touches this file, and the changes cannot be automatically merged.

Fix:

This is not a patch defect — the PR commits themselves are clean. The conflict arises from the CI integration workflow merging multiple topic branches together. The PR author has two options:
Rebase the PR branch onto the latest integration baseline to resolve conflicts locally:
git fetch origin
git rebase origin/qcom-next  # or the appropriate base branch
# Resolve conflicts in Makefile and kodiak-el2.dtso
git add arch/arm64/boot/dts/qcom/Makefile arch/arm64/boot/dts/qcom/kodiak-el2.dtso
git rebase --continue
git push --force-with-lease
Wait for the conflicting topic branch to merge first, then re-trigger CI. The conflict may resolve itself once the integration baseline advances.
Reproduce locally:
# Clone the kernel-topics repo
git clone https://github.com/qualcomm-linux/kernel-topics.git
cd kernel-topics
git fetch origin pull/2159/head:pr-2159
git checkout pr-2159

# Attempt to merge into the integration baseline
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c7896917e523946138ce654d4b6c1e33d5a8f4e5
git rebase c7896917e523946138ce654d4b6c1e33d5a8f4e5
# Conflicts will appear here
✅ tag-check

Status: PASS

Both commits in the PR have valid subject-line prefixes:

Commit 1: FROMLIST: arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA memory region for AudioReach

Commit 2: FROMLIST: arm64: dts: qcom: kodiak-el2: add LPASS AML IOMMU SIDs for eNPU audio offload

Both commits include Link: trailers pointing to the upstream lore.kernel.org submission.

Verdict

No checker defects found. The CI failure is due to merge conflicts during the integration step, not issues with the patch content itself. The PR commits are properly formatted with FROMLIST: prefixes and upstream links.

Recommended action: Rebase the PR branch onto the latest integration baseline to resolve the Makefile and kodiak-el2.dtso conflicts, then re-trigger CI. Once the automerge succeeds, all checkers will run and provide validation results.

…on for AudioReach Add a dynamic CMA pool for the AudioReach audio stack on qcs6490-rb3gen2. Use alloc-ranges to let the kernel pick a free address rather than a fixed reg, which avoided EFI-reserved memory at 0x9cc00000. Add 4 MiB alignment to satisfy CMA_MIN_ALIGNMENT_BYTES. Wire memory-region into q6apmdai so the q6apm-audio-mem-cma child driver can attach the pool. Based on the equivalent implementation for lemans (SA8295P). Signed-off-by: Sairamreddy Bojja <sbojja@qti.qualcomm.com>

…NPU audio offload Add the three LPASS AML stream IDs to remoteproc_adsp in the kodiak EL2 overlay so that the q6apm-audio-mem-cma child driver can obtain an IOMMU domain via iommu_get_domain_for_dev(dev). Without these SIDs the CMA device has no IOMMU domain (has_iommu=0) and IOCTL_MAP_HYP_ASSIGN falls through to the SCM path, which fails on RB3Gen2 because Gunyah owns the SMMU. SID assignments: 0x1860 / mask 0x9 — ML Activation + Weights + Write streams 0x1862 / mask 0x1 — ML Config + Bias streams 0x1864 / mask 0x0 — ML Normalized Data stream Signed-off-by: Sairamreddy Bojja <sbojja@qti.qualcomm.com>

qlijarvis · 2026-10-30T07:30:57Z

🔨 Build Failure Analysis — PR #2159

PR: #2159
Build run: https://github.com/qualcomm-linux/kernel-config/actions/runs/41018331453

#	Error	File:Line	PR-introduced?	Root Cause
1	ERROR (phandle_references): Reference to non-existent node or label "audio_cma_mem"	arch/arm64/boot/dts/qcom/qcs6490-audioreach.dtsi:100	Yes	The PR adds a reference to `audio_cma_mem` in the shared include file `qcs6490-audioreach.dtsi`, but only defines the node in `qcs6490-rb3gen2.dts`. Other boards that include `qcs6490-audioreach.dtsi` (qcm6490-idp, qcs6490-radxa-dragon-q6a) fail because they don't have the `audio_cma_mem` node.

Verdict

All build errors are introduced by this PR. The error affects 2 device tree blobs (qcm6490-idp.dtb and qcs6490-radxa-dragon-q6a.dtb) in both standard and RT builds, causing 4 total build failures.

📎 Detailed analysis: Full report

qlijarvis · 2026-10-30T07:35:45Z

🔨 Build Failure Analysis — PR #2159

PR: #2159
Build run: https://github.com/qualcomm-linux/kernel-config/actions/runs/41018331453

#	Error	File:Line	PR-introduced?	Root Cause
1	ERROR (phandle_references): Reference to non-existent node or label "audio_cma_mem"	arch/arm64/boot/dts/qcom/qcs6490-audioreach.dtsi:97-101	Yes	PR adds `memory-region = <&audio_cma_mem>` to shared .dtsi but only defines `audio_cma_mem` in qcs6490-rb3gen2.dts, breaking qcm6490-idp.dtb and qcs6490-radxa-dragon-q6a.dtb builds

Verdict

1 device tree compilation error introduced by this PR affecting 2 DTB targets (qcm6490-idp.dtb and qcs6490-radxa-dragon-q6a.dtb).

📎 Detailed analysis: Full report

qlijarvis · 2026-10-30T07:57:17Z

PR #2159 — validate-patch

PR: #2159

Verdict	Issues	Detailed Report
⚠️	3	Full report

Final Summary

Lore link present: No — PENDING: prefix; no lore link expected for work-in-progress commits
Lore link matches PR commits: N/A — no lore link to compare against
Upstream patch status: N/A — not posted upstream; marked as PENDING (work-in-progress)
PR present in qcom-next/topics: Fail - 2/2 commit(s) are missing from both qcom-next and topics

Verdict: ⚠️ — click to expand

🔍 Patch Validation

PR: #2159
Upstream commit: N/A (PENDING: prefix — not yet posted upstream)
Verdict: ⚠️ PARTIAL

Commit 1/2: PENDING: arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA

Commit Message

Check Status Note

Subject matches upstream N/A No upstream posting yet

Body preserves rationale ✅ Clear explanation of CMA pool purpose and implementation choices

Fixes tag present/correct N/A Not a fix; new feature

Authorship preserved ✅ Signed-off-by present

Backport note (if applicable) N/A Not a backport

Diff

File Status Notes

arch/arm64/boot/dts/qcom/qcs6490-audioreach.dtsi ✅ Adds memory-region property to q6apmdai

arch/arm64/boot/dts/qcom/qcs6490-rb3gen2.dts ✅ Adds audio_cma_mem CMA pool with proper alignment

Commit 2/2: PENDING: arm64: dts: qcom: kodiak-el2: add LPASS AML IOMMU SIDs

Commit Message

Check Status Note

Subject matches upstream N/A No upstream posting yet

Body preserves rationale ✅ Detailed explanation of SID assignments and why they're needed

Fixes tag present/correct N/A Not a fix; new feature

Authorship preserved ✅ Signed-off-by present

Backport note (if applicable) N/A Not a backport

Diff

File Status Notes

arch/arm64/boot/dts/qcom/kodiak-el2.dtso ✅ Adds three IOMMU stream IDs with correct masks

Issues

Commit 1/2:

⚠️ Missing upstream posting: This appears to be upstreamable board-specific DTS configuration. The commit message references "equivalent implementation for lemans (SA8295P)", suggesting similar patches may already be upstream. Consider posting to the mailing list or changing prefix to QCLINUX: if this is intentionally vendor-only.

Commit 2/2:

⚠️ Missing upstream posting: Board-specific IOMMU configuration that appears upstreamable. The detailed technical explanation in the commit message is suitable for upstream review. Consider posting to devicetree@vger.kernel.org and linux-arm-msm@vger.kernel.org.

Both commits:

⚠️ Not present in qcom-next or topics: According to integration_presence_report.md, both commits are missing from qcom-next and all topic branches. For PENDING: commits this may be expected if they're truly work-in-progress, but it raises the question of whether they should be in a topic branch first before merging.

Verdict

PARTIAL — The commits are technically correct and well-documented, but the PENDING: prefix usage needs clarification:

If these are genuinely work-in-progress and will be posted upstream soon, the PENDING: prefix is appropriate, but they should ideally go through a topic branch first.

If these are vendor-specific configurations not intended for upstream, they should use QCLINUX: prefix instead.

If similar functionality already exists upstream (as suggested by the lemans reference in commit 1/2), consider using UPSTREAM: or BACKPORT: with appropriate lore links.

Recommendation: Before merging, clarify the upstream intent. If upstreaming is planned, post to the mailing list and update to FROMLIST: with lore links. If vendor-only, change to QCLINUX:. If based on existing upstream work, add appropriate lore references.

Final Summary

Lore link present: No — PENDING: prefix; no lore link expected for work-in-progress commits

Lore link matches PR commits: N/A — no lore link to compare against

Upstream patch status: N/A — not posted upstream; marked as PENDING (work-in-progress)

PR present in qcom-next/topics: Fail — 2/2 commits are missing from both qcom-next and topics (per integration_presence_report.md)

Deterministic Integration Presence

Integration Presence Report

This report is generated by Jarvis before validate-patch runs.
It is the authoritative source for whether PR changes are already present
in qcom-next or in the kernel topic branches.

Kernel repo: /local/mnt/workspace/sgaud/Qgenie/image_pipeline/kernel
qcom-next ref: c7896917e523946138ce654d4b6c1e33d5a8f4e5
topics remote: topics -> https://github.com/qualcomm-linux/kernel-topics
topics fetch: fetched

Commit Subject qcom-next topics Final

1/2 [PATCH 1/2] PENDING: arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA missing - no subject, patch-id, or full tree-content match found missing - no subject, patch-id, or full tree-content match found missing

2/2 [PATCH 2/2] PENDING: arm64: dts: qcom: kodiak-el2: add LPASS AML missing - no subject, patch-id, or full tree-content match found missing - no subject, patch-id, or full tree-content match found missing

Final Status

overall_status: FAIL
present_commits: 0/2
partial_commits: 0/2
missing_commits: 2/2
topics_checked_for_commits: 2/2
final_summary: PR present in qcom-next/topics: Fail - 2/2 commit(s) are missing from both qcom-next and topics

qlijarvis · 2026-10-30T07:57:20Z

PR #2159 — checker-log-analyzer

PR: #2159
Checker run: https://github.com/qualcomm-linux/kernel-config/actions/runs/41018330899

Checker	Result	Summary
Checker	Result	Summary
checkpatch	⏭️	Skipped - automerge failed
dt-binding-check	⏭️	Skipped - automerge failed
dtb-check	⏭️	Skipped - automerge failed
sparse-check	⏭️	Skipped - automerge failed
check-uapi-headers	⏭️	Skipped - automerge failed
check-patch-compliance	⏭️	Skipped - automerge failed
tag-check	N/A	Not applicable - PR targets topic branch
qcom-next-check	⚠️	Both commits use `PENDING:` prefix

Detailed report: Full report

Checker analysis — click to expand

🤖 CI Checker Analysis (checker-log-analyzer)

PR: #2159 - PENDING: arm64: dts: qcom: qcs6490-rb3gen2/kodiak-el2 audio changes
Source: https://github.com/qualcomm-linux/kernel-config/actions/runs/41018330899

Checker Result Summary

checkpatch ⏭️ Skipped - automerge failed

dt-binding-check ⏭️ Skipped - automerge failed

dtb-check ⏭️ Skipped - automerge failed

sparse-check ⏭️ Skipped - automerge failed

check-uapi-headers ⏭️ Skipped - automerge failed

check-patch-compliance ⏭️ Skipped - automerge failed

tag-check N/A Not applicable - PR targets topic branch

qcom-next-check ⚠️ Both commits use PENDING: prefix

❌ Automerge Failure (Pre-Checker Stage)

Root cause: Merge conflicts during integration of topic/tech/all/dt/qcs6490 with baseline c7896917e523946138ce654d4b6c1e33d5a8f4e5

Failure details:
Merging topic branch: topic/tech/all/dt/qcs6490
Auto-merging arch/arm64/boot/dts/qcom/Makefile
CONFLICT (content): Merge conflict in arch/arm64/boot/dts/qcom/Makefile
Auto-merging arch/arm64/boot/dts/qcom/kodiak-el2.dtso
CONFLICT (add/add): Merge conflict in arch/arm64/boot/dts/qcom/kodiak-el2.dtso
Automatic merge failed; fix conflicts and then commit the result.
Analysis:

arch/arm64/boot/dts/qcom/Makefile — Content conflict
The PR likely adds new DTB/DTSO entries that conflict with other changes in the topic branch.

arch/arm64/boot/dts/qcom/kodiak-el2.dtso — Add/add conflict
Both the PR (commit 2/2) and the topic branch modify kodiak-el2.dtso. The PR adds LPASS AML IOMMU SIDs to &remoteproc_adsp, while the topic branch likely has overlapping changes to the same overlay.

Fix:

The PR author needs to rebase on the latest topic/tech/all/dt/qcs6490 branch and resolve conflicts:
# Fetch the latest topic branch
git fetch topics topic/tech/all/dt/qcs6490

# Rebase the PR branch
git rebase topics/topic/tech/all/dt/qcs6490

# Resolve conflicts in:
#   - arch/arm64/boot/dts/qcom/Makefile
#   - arch/arm64/boot/dts/qcom/kodiak-el2.dtso

# For Makefile: merge the DTB/DTSO entries in sorted order
# For kodiak-el2.dtso: combine the iommus property changes

git add <resolved-files>
git rebase --continue
Reproduce locally:
git clone https://github.com/qualcomm-linux/kernel-topics
cd kernel-topics
git fetch origin pull/2159/head:pr2159
git checkout pr2159
git fetch topics topic/tech/all/dt/qcs6490
git rebase topics/topic/tech/all/dt/qcs6490
⚠️ Subject Prefix Notice

Both commits use the PENDING: prefix:

Commit 1/2: PENDING: arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA memory region for AudioReach

Commit 2/2: PENDING: arm64: dts: qcom: kodiak-el2: add LPASS AML IOMMU SIDs for eNPU audio offload

Note: PENDING: is a vendor-internal prefix used for work-in-progress commits. While this is acceptable for topic branches, be aware that:

check-patch-compliance will fail for PENDING: commits (known limitation)

These commits should eventually be posted upstream and re-prefixed as FROMLIST: or UPSTREAM:

Verdict

Cannot merge — The PR has merge conflicts with the target topic branch topic/tech/all/dt/qcs6490. The author must rebase and resolve conflicts before the checkers can run. Once conflicts are resolved, re-trigger CI to validate the patches.

qti-sbojja changed the title ~~CMA-memory-to-offload-FNN-PLUS-module-to-eNPU- #1023~~ CMA-memory-to-offload-FNN-PLUS-module-to-eNPU Oct 29, 2026

qcomlnxci requested review from a team, aiquny, Nicolas Dechesne (ndechesne) and Trilok Soni (trsoni) and removed request for a team October 29, 2026 11:21

qti-sbojja force-pushed the audio-ML-K branch from d5cad51 to 805b198 Compare October 29, 2026 11:23

qcomlnxci requested a review from a team October 29, 2026 11:25

qti-sbojja force-pushed the audio-ML-K branch from 805b198 to 2960c50 Compare October 30, 2026 05:20

qti-sbojja added 2 commits October 30, 2026 12:37

qti-sbojja force-pushed the audio-ML-K branch from 2960c50 to cd85a67 Compare October 30, 2026 07:08

Conversation

qti-sbojja commented Oct 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

qlijarvis commented Oct 29, 2026

🔨 Build Failure Analysis — PR #2159

Verdict

Uh oh!

qlijarvis commented Oct 29, 2026

🔨 Build Failure Analysis — PR #2159

Verdict

Uh oh!

qlijarvis commented Oct 29, 2026

PR #2159 — validate-patch

Final Summary

🔍 Patch Validation

Commit Message

Diff

Upstream Patch Status

Issues

Verdict

Final Summary

Deterministic Integration Presence

Integration Presence Report

Final Status

Uh oh!

qlijarvis commented Oct 29, 2026

PR #2159 — checker-log-analyzer

🤖 CI Checker Analysis (checker-log-analyzer)

❌ Automerge Failure (Pre-Checker Stage)

✅ tag-check (Manual Verification)

📋 Patch Content Summary

Verdict

Uh oh!

qlijarvis commented Oct 30, 2026

🔨 Build Failure Analysis — PR #2159

Verdict

Uh oh!

qlijarvis commented Oct 30, 2026

🔨 Build Failure Analysis — PR #2159

Verdict

Uh oh!

qlijarvis commented Oct 30, 2026

PR #2159 — validate-patch

Final Summary

🔍 Patch Validation

Commit Message

Diff

Issues

Upstream Patch Status

Dependency Check

qcom-next/topics Presence

Verdict

Final Summary

Deterministic Integration Presence

Integration Presence Report

Final Status

Uh oh!

qlijarvis commented Oct 30, 2026

PR #2159 — checker-log-analyzer

🤖 CI Checker Analysis (checker-log-analyzer)

⚠️ Automerge Failure (Pre-Checker Stage)

✅ tag-check

Verdict

Uh oh!

qlijarvis commented Oct 30, 2026

🔨 Build Failure Analysis — PR #2159

Verdict

Uh oh!

qlijarvis commented Oct 30, 2026

🔨 Build Failure Analysis — PR #2159

Verdict

Uh oh!

qlijarvis commented Oct 30, 2026

PR #2159 — validate-patch

Final Summary

🔍 Patch Validation

Commit 1/2: PENDING: arm64: dts: qcom: qcs6490-rb3gen2: add audio CMA

Commit Message

Diff

qti-sbojja commented Oct 29, 2026 •

edited

Loading