Audio ML support for Lemans and Monaco by qti-sbojja · Pull Request #840 · qualcomm-linux/kernel

qti-sbojja · 2026-07-18T05:42:12Z

The audio ML (machine-learning) offload use case requires a contiguous,
physically addressable buffer shared with the audio DSP/SPF, and issues
DMA transactions through additional SMMU stream IDs that are not
covered by the existing ADSP mapping.

Add a dedicated reusable shared-dma-pool CMA region (16 MiB, 4 MiB
aligned) and wire it to the q6apm DAIs node via memory-region. A
dedicated pool guarantees the alignment and contiguity the DSP expects
and isolates these allocations from the default CMA region.

Under the EL2 (Gunyah/hypervisor) configuration the SMMU is fully
enforcing, so the ML transactions are otherwise blocked and faulted:

arm-smmu 15000000.iommu: Blocked unknown Stream ID 0x3060
arm-smmu 15000000.iommu: Blocked unknown Stream ID 0x3062

Add the ML-related stream IDs (0x3060 mask 0x9, 0x3062 mask 0x1) to the
remoteproc_adsp iommus property in the EL2 overlay so these buffers are
translated by the SMMU instead of being rejected.

CRs-Fixed: 4613255

qlijarvis · 2026-07-18T06:32:32Z

PR #840 — validate-patch

PR: #840

Verdict	Issues	Detailed Report
✅	0	Full report

Final Summary

Lore link present: Yes — all three commits include correct Link: tags pointing to lore.kernel.org message-IDs
Lore link matches PR commits: Yes — diff content is identical to lore patches; line number shifts are legitimate context differences between upstream and qcom tree bases
Upstream patch status: ⏳ Decision Pending — patches posted to linux-arm-msm@vger.kernel.org on 2026-07-17 (v2 series); only automated AI review replies present in thread (no maintainer ACK/NAK yet); last activity 2026-07-17 10:13 UTC
PR present in qcom-next/topics: Yes - all 3 commit(s) are present in qcom-next or topics

Verdict: ✅ — click to expand

🔍 Patch Validation

PR: #840 - FROMLIST: dt-bindings: remoteproc: qcom,pas-common: allow up to 3 iommus (+ lemans/monaco audio ML offload)
Upstream commits:

Patch 1/3: https://lore.kernel.org/all/20260717095825.3962666-2-pratyush.meduri@oss.qualcomm.com/

Patch 2/3: https://lore.kernel.org/all/20260717095825.3962666-3-pratyush.meduri@oss.qualcomm.com/

Patch 3/3: https://lore.kernel.org/all/20260717095825.3962666-4-pratyush.meduri@oss.qualcomm.com/
Verdict: ✅ PASS

Commit Message

Check Status Note

Subject matches upstream ✅ All three subjects match lore patches exactly (with FROMLIST: prefix added)

Body preserves rationale ✅ All commit bodies faithfully preserve the upstream rationale and technical details

Fixes tag present/correct N/A No Fixes tags in upstream patches

Authorship preserved ✅ All three commits correctly preserve From: Pratyush Meduri <pratyush.meduri@oss.qualcomm.com> and Signed-off-by: from lore

Backport note (if applicable) N/A These are FROMLIST: commits (posted to mailing list, not yet merged), not backports

Diff

File Status Notes

Documentation/devicetree/bindings/remoteproc/qcom,pas-common.yaml ✅ Identical change (iommus: maxItems 1→3, minItems 1 added); line numbers differ (33 vs 45) due to different tree base

arch/arm64/boot/dts/qcom/lemans-el2.dtso ✅ Identical change (adds SMMU stream IDs 0x3060/0x3062); line 19 in both

arch/arm64/boot/dts/qcom/lemans.dtsi ✅ Identical change (adds audio_cma_mem pool + memory-region); line numbers differ (624 vs 804, 8142 vs 7786) due to different tree base

arch/arm64/boot/dts/qcom/monaco-el2.dtso ✅ Identical change (adds SMMU stream IDs 0x2060/0x2062); line 17 in lore vs 20 in PR (context-only delta)

arch/arm64/boot/dts/qcom/monaco.dtsi ✅ Identical change (adds audio_cma_mem pool + memory-region); line numbers differ (729 vs 789, context varies) due to different tree base

Verdict

✅ Merge as-is — All three commits are faithful representations of the upstream lore patches. The diff content is identical; line number and file hash differences are expected and correct for a FROMLIST: integration into the Qualcomm kernel tree. Commit messages, authorship, and technical rationale are all preserved correctly.

Final Summary

Lore link present: Yes — all three commits include correct Link: tags pointing to lore.kernel.org message-IDs

Lore link matches PR commits: Yes — diff content is identical to lore patches; line number shifts are legitimate context differences between upstream and qcom tree bases

Upstream patch status: ⏳ Decision Pending — patches posted to linux-arm-msm@vger.kernel.org on 2026-07-17 (v2 series); only automated AI review replies present in thread (no maintainer ACK/NAK yet); last activity 2026-07-17 10:13 UTC

PR present in qcom-next/topics: Yes — integration_presence_report.md confirms all 3 commits present: commit 1/3 in qcom-next, commits 2/3 and 3/3 in topics (exact patch-id match for 3/3 at 251e0b59a16264316858e927e7fecf58ec1a3e06)

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: a5cf3debd8c3c660711ad586ad4bb84e9ca42635
topics remote: topics -> https://github.com/qualcomm-linux/kernel-topics
topics fetch: fetched

Commit Subject qcom-next topics Final

1/3 [PATCH 1/3] FROMLIST: dt-bindings: remoteproc: qcom,pas-common: allow present - all checked added lines are present skipped - not checked because qcom-next already contains the change present

2/3 [PATCH 2/3] FROMLIST: arm64: dts: qcom: lemans: enable audio ML missing - no subject, patch-id, or full tree-content match found present - all checked added lines are present present

3/3 [PATCH 3/3] FROMLIST: arm64: dts: qcom: monaco: enable audio ML missing - no subject, patch-id, or full tree-content match found present - exact patch-id match at 251e0b59a16264316858e927e7fecf58ec1a3e06 present

Final Status

overall_status: PASS
present_commits: 3/3
partial_commits: 0/3
missing_commits: 0/3
topics_checked_for_commits: 2/3
final_summary: PR present in qcom-next/topics: Yes - all 3 commit(s) are present in qcom-next or topics

qlijarvis · 2026-07-18T06:32:33Z

PR #840 — checker-log-analyzer

PR: #840
Checker run: https://github.com/qualcomm-linux/kernel-config/actions/runs/29632759870

Checker	Result	Summary
Checker	Result	Summary
checkpatch	✅	All 3 commits passed
dt-binding-check	✅	Binding schema validation passed
dtb-check	❌	Missing required `qcom,vmid` property
sparse-check	⏭️	Skipped (no C/H changes)
check-uapi-headers	⏭️	Skipped (no UAPI changes)
check-patch-compliance	✅	All commits have valid `FROMLIST:` prefix and Link
tag-check	✅	All commits have valid `FROMLIST:` prefix

Detailed report: Full report

Checker analysis — click to expand

🤖 CI Checker Analysis (checker-log-analyzer)

PR: #840 - FROMLIST: dt-bindings: remoteproc: qcom,pas-common: allow up to 3 iommus
Source: https://github.com/qualcomm-linux/kernel-config/actions/runs/29632759870
Target branch: qcom-6.18.y

Checker Result Summary

checkpatch ✅ All 3 commits passed

dt-binding-check ✅ Binding schema validation passed

dtb-check ❌ Missing required qcom,vmid property

sparse-check ⏭️ Skipped (no C/H changes)

check-uapi-headers ⏭️ Skipped (no UAPI changes)

check-patch-compliance ✅ All commits have valid FROMLIST: prefix and Link

tag-check ✅ All commits have valid FROMLIST: prefix

❌ dtb-check

Root cause: The PR adds memory-region property to the q6apm-dais node but does not add the required qcom,vmid property, violating the binding schema dependency rule.

Failure details:
/opt/actions-runner/_work/kernel-config/kernel-config/kernel/temp-out/arch/arm64/boot/dts/qcom/lemans-evk.dtb: 
  dais (qcom,q6apm-dais): 'qcom,vmid' is a dependency of 'memory-region'
	from schema $id: http://devicetree.org/schemas/sound/qcom,q6apm-dai.yaml#

/opt/actions-runner/_work/kernel-config/kernel-config/kernel/temp-out/arch/arm64/boot/dts/qcom/lemans-evk-el2.dtb: 
  dais (qcom,q6apm-dais): 'qcom,vmid' is a dependency of 'memory-region'
	from schema $id: http://devicetree.org/schemas/sound/qcom,q6apm-dai.yaml#

/opt/actions-runner/_work/kernel-config/kernel-config/kernel/temp-out/arch/arm64/boot/dts/qcom/monaco-evk.dtb: 
  dais (qcom,q6apm-dais): 'qcom,vmid' is a dependency of 'memory-region'
	from schema $id: http://devicetree.org/schemas/sound/qcom,q6apm-dai.yaml#
Additionally, there is a secondary error:
service@1 (qcom,q6apm): Unevaluated properties are not allowed ('dais' was unexpected)
	from schema $id: http://devicetree.org/schemas/sound/qcom,q6apm.yaml#
Analysis:

The binding schema Documentation/devicetree/bindings/sound/qcom,q6apm-dai.yaml defines a dependency rule:
dependencies:
  memory-region: ['qcom,vmid']
This means: if the memory-region property is present in a qcom,q6apm-dais node, then the qcom,vmid property must also be present.

The PR adds memory-region = <&audio_cma_mem>; to the dais node in:

arch/arm64/boot/dts/qcom/lemans.dtsi:807

arch/arm64/boot/dts/qcom/monaco.dtsi (similar location)

But does not add the required qcom,vmid property.

Fix:

Add the qcom,vmid property to the dais node wherever memory-region is added:
 q6apmdai: dais {
 	compatible = "qcom,q6apm-dais";
 	iommus = <&apps_smmu 0x3001 0x0>;
 	memory-region = <&audio_cma_mem>;
+	qcom,vmid = <VMID_HLOS>;
 };
The VMID_HLOS constant is typically defined in include/dt-bindings/firmware/qcom,scm.h or a similar header. Check the binding documentation and existing usage in the tree to determine the correct VMID value for the audio ML offload use case.

Reproduce locally:
make -j$(nproc) O=out defconfig
make -j$(nproc) O=out CHECK_DTBS=y arch/arm64/boot/dts/qcom/lemans-evk.dtb
make -j$(nproc) O=out CHECK_DTBS=y arch/arm64/boot/dts/qcom/monaco-evk.dtb
Verdict

1 blocker must be fixed before merge:

The dtb-check failure is a genuine schema violation introduced by the PR. The qcom,vmid property must be added to the dais node wherever memory-region is added, as required by the qcom,q6apm-dai.yaml binding schema dependency rule.

All other checkers passed successfully.

The ADSP PAS on several Qualcomm SoCs (e.g. SA8775P/lemans, QCS8300, Monaco) needs more than a single SMMU stream ID for the audio ML (machine-learning) offload use case. In addition to the primary ADSP stream ID, the DSP/SPF issues DMA transactions through extra stream IDs that are not covered by the existing single-entry mapping. Relax the iommus constraint to allow between 1 and 3 entries so the additional ML offload stream IDs can be described, fixing dtb_check schema violations on device trees that add them. Remove the dependency requiring qcom,vmid whenever a memory-region property is present in the q6apm-dai binding. The memory region itself does not inherently require VMID configuration and the dependency unnecessarily restricts valid DT descriptions. Link: https://lore.kernel.org/all/20260717095825.3962666-2-pratyush.meduri@oss.qualcomm.com/ Signed-off-by: Pratyush Meduri <pratyush.meduri@oss.qualcomm.com>

…d SMMU mappings The audio ML (machine-learning) offload use case requires a contiguous, physically addressable buffer shared with the audio DSP/SPF, and issues DMA transactions through additional SMMU stream IDs that are not covered by the existing ADSP mapping. Add a dedicated reusable shared-dma-pool CMA region (16 MiB, 4 MiB aligned) and wire it to the q6apm DAIs node via memory-region. A dedicated pool guarantees the alignment and contiguity the DSP expects and isolates these allocations from the default CMA region. Under the EL2 (Gunyah/hypervisor) configuration the SMMU is fully enforcing, so the ML transactions are otherwise blocked and faulted: arm-smmu 15000000.iommu: Blocked unknown Stream ID 0x3060 arm-smmu 15000000.iommu: Blocked unknown Stream ID 0x3062 Add the ML-related stream IDs (0x3060 mask 0x9, 0x3062 mask 0x1) to the remoteproc_adsp iommus property in the EL2 overlay so these buffers are translated by the SMMU instead of being rejected. Link: https://lore.kernel.org/all/20260717095825.3962666-3-pratyush.meduri@oss.qualcomm.com/ Signed-off-by: Pratyush Meduri <pratyush.meduri@oss.qualcomm.com>

…d SMMU mappings The audio ML (machine-learning) offload use case requires a contiguous, physically addressable buffer shared with the audio DSP/SPF, and issues DMA transactions through additional SMMU stream IDs that are not covered by the existing ADSP mapping. Add a dedicated reusable shared-dma-pool CMA region (16 MiB, 4 MiB aligned) and wire it to the q6apm DAIs node via memory-region. A dedicated pool guarantees the alignment and contiguity the DSP expects and isolates these allocations from the default CMA region. Under the EL2 (Gunyah/hypervisor) configuration the SMMU is fully enforcing, so the ML transactions are otherwise blocked and faulted: arm-smmu 15000000.iommu: Blocked unknown Stream ID 0x2060 arm-smmu 15000000.iommu: Blocked unknown Stream ID 0x2062 Add the ML-related stream IDs (0x3060 mask 0x9, 0x3062 mask 0x1) to the remoteproc_adsp iommus property in the EL2 overlay so these buffers are translated by the SMMU instead of being rejected. Link: https://lore.kernel.org/all/20260717095825.3962666-4-pratyush.meduri@oss.qualcomm.com/ Signed-off-by: Pratyush Meduri <pratyush.meduri@oss.qualcomm.com>

qswat-orbit-external · 2026-07-20T09:47:58Z

Merge Check Failed: CR Not Eligible for Merge

CR 4613255 is not eligible for merge.

The parent software image for kernel.qli.2.0 is not development complete.

Entity: kernel.qli.2.0
CR: 4613255
Reason: CR_CANNOT_MERGE

Please ensure the CR passes both CCT (ComponentChangeTasks) and ICT (Integration Change Tasks) validations.

qswat-orbit-external · 2026-07-20T09:48:11Z

Merge Check Failed: CR Not Eligible for Merge

CR 4613255 is not eligible for merge.

The parent software image for kernel.qli.2.0 is not development complete.

Entity: kernel.qli.2.0
CR: 4613255
Reason: CR_CANNOT_MERGE

Please ensure the CR passes both CCT (ComponentChangeTasks) and ICT (Integration Change Tasks) validations.

qswat-orbit-external · 2026-07-20T09:50:54Z

Merge Check Failed: CR Not Eligible for Merge

CR 4613255 is not eligible for merge.

The parent software image for kernel.qli.2.0 is not development complete.

Entity: kernel.qli.2.0
CR: 4613255
Reason: CR_CANNOT_MERGE

Please ensure the CR passes both CCT (ComponentChangeTasks) and ICT (Integration Change Tasks) validations.

qti-sbojja requested review from a team, Mukesh Ojha (mukeshojha-linux), Rajendra (rrnayak) and Salendarsingh Gaud (sgaud-quic) July 18, 2026 05:42

Pratyush Meduri added 3 commits July 20, 2026 12:09

qti-sbojja force-pushed the audio-ML branch from 13201cf to 7b19957 Compare July 20, 2026 06:41

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Audio ML support for Lemans and Monaco#840

Audio ML support for Lemans and Monaco#840
qti-sbojja wants to merge 3 commits into
qualcomm-linux:qcom-6.18.yfrom
qti-sbojja:audio-ML

qti-sbojja commented Jul 18, 2026 •

edited

Loading

Uh oh!

qlijarvis commented Jul 18, 2026

🔍 Patch Validation

Commit Message

Diff

Verdict

Final Summary

Deterministic Integration Presence

Integration Presence Report

Final Status

Uh oh!

qlijarvis commented Jul 18, 2026

🤖 CI Checker Analysis (checker-log-analyzer)

❌ dtb-check

Verdict

Uh oh!

qswat-orbit-external Bot commented Jul 20, 2026

Uh oh!

qswat-orbit-external Bot commented Jul 20, 2026

Uh oh!

qswat-orbit-external Bot commented Jul 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

qti-sbojja commented Jul 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

qlijarvis commented Jul 18, 2026

PR #840 — validate-patch

Final Summary

🔍 Patch Validation

Commit Message

Diff

Verdict

Final Summary

Deterministic Integration Presence

Integration Presence Report

Final Status

Uh oh!

qlijarvis commented Jul 18, 2026

PR #840 — checker-log-analyzer

🤖 CI Checker Analysis (checker-log-analyzer)

❌ dtb-check

Verdict

Uh oh!

qswat-orbit-external Bot commented Jul 20, 2026

Uh oh!

qswat-orbit-external Bot commented Jul 20, 2026

Uh oh!

qswat-orbit-external Bot commented Jul 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

qti-sbojja commented Jul 18, 2026 •

edited

Loading