feat: adding support for amdama (supernova) gpus release 1.5

[mipresmsft]

Updates node provisioning + e2e validation support for AMD AMA (Supernova / MA35D) GPUs release 1.5 in the Linux CSE flow, primarily targeting Azure Linux v3. New naming scheme should also be forward compatible with future driver/SW releases from AMD.

Changes:

Update AMD AMA install logic to support new file naming convention, separated driver/FW packages.
Add ShellSpec test code.

[Copilot]

Pull request overview

Updates Linux CSE + e2e coverage to support provisioning/validation for AMD AMA (Supernova / MA35D) GPUs release 1.5, primarily on Azure Linux v3, and adjusts e2e firewall rules accordingly.

Changes:

• Updates Azure Linux MA35D provisioning logic to discover/install AMD AMA driver/firmware and derive matching “core” package coordinates.
• Adds/updates MA35D e2e validation and simplifies the scenario’s system pool SKU usage (defaults).
• Removes download.microsoft.com from the e2e firewall allowlist.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

File	Description
parts/linux/cloud-init/artifacts/cse_config.sh	Reworks AMD AMA install flow (repoquery-based driver selection, firmware install, core package URL construction).
e2e/scenario_test.go	Tweaks AzureLinuxV3 MA35D scenario config (removes explicit system pool SKU override).
e2e/aks_model.go	Removes download.microsoft.com from Azure Firewall application rules used by e2e.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 4 comments.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 5 comments.

[djsly]

AgentBaker Linux gate detective: CIS regression

Run: https://msazure.visualstudio.com/CloudNativeCompute/_build/results?buildId=168488433
Failed job/task: build2204gen2containerd / Test, Scan, and Cleanup
Wiki signature: linux-vhd-prgate-cis-ubuntu2204-gen2-containerd-6131-logfiles

Detective summary: The build is partially succeeded because CIS scan comparison logged one regression: rule 6.1.3.1 (Ensure access to all logfiles has been configured) changed pass -> fail on Ubuntu 22.04 Gen2 containerd. The test and scan scripts exited 0 after uploading cis-report-l1/l2 and HTML reports, so this is a VHD/CIS compliance signal rather than an E2E failure.

Likely cause: Shared Ubuntu 22.04 VHD/CIS image-state drift or a generated logfile permission/access mismatch. This same CIS rule appeared earlier today on an unrelated PR in the ARM64 lane, so PR causality is currently unlikely.

Confidence: Medium-high. Evidence comes from build status, timeline/task issue, task log regression details, uploaded CIS report names, passing test-result preview, and recurrence across unrelated PRs/lanes.

Strongest alternative theory: This PR's AMD AMA CSE change somehow creates a non-compliant logfile in the Ubuntu VHD scan path. Less likely because the changed code is MA35D/Azure Linux AMD package install logic, while the failure is a generic Ubuntu 22.04 logfile-access CIS rule and recurred on an unrelated PR.

Recommended next action / owner: Linux VHD/CIS build owner should inspect the uploaded CIS reports for build 168488433 to identify the exact logfile path, owner/group, and mode, then decide whether to fix image generation permissions or update the CIS baseline for expected package drift.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

[aks-node-assistant]

AgentBaker Linux gate detective

Run: https://msazure.visualstudio.com/CloudNativeCompute/_build/results?buildId=168752702
Failed job/stage/task: $(System.Collections.Hashtable.job)
First failing step/test: $(System.Collections.Hashtable.first)

RCA: The VHD image built, but post-build VHD content test failed inside the test VM while cloning AgentBaker: git fetch of refs/pull/8643/merge returned remote ref not found. This is a VHD content-test harness / PR merge-ref availability issue, not a Packer build failure. Secondary: Ubuntu 22.04 CIS 6.1.3.1 pass->fail matches repair item #38501652.

Confidence: High for the primary signature. Corroborated by timeline/status, focused failed logs, associated changes, and the flakiness wiki before publishing.

Strongest alternative: the PR changes broke MA35D validation. Less likely because the test aborts during AgentBaker checkout before content validation.

Recommended owner/action: VHD content-test/build harness owner: preserve source checkout or use a commit SHA/ref available from the test VM instead of relying on transient PR merge refs.

Wiki signature: $(System.Collections.Hashtable.sig) (source of truth)

[aks-node-assistant]

AgentBaker Linux gate detective

Run: https://msazure.visualstudio.com/CloudNativeCompute/_build/results?buildId=168523974
Failed job/stage/task: $job
RCA: Known VHD content-test harness failure: post-build content test tries to fetch a transient refs/pull/*/merge ref from inside the test VM and GitHub returns remote ref not found before content validation.

Confidence: High. Corroborated by prior focused logs, repeated unrelated PRs, and the wiki source of truth.

Strongest alternative: the PR product change caused content-test failure; less likely because the test aborts while cloning AgentBaker before validation.

Recommended owner/action: E2E/build harness owner should follow the linked wiki signature and repair item if present; rerun after the infra/harness issue is cleared.

Wiki signature: $sig (source of truth)