KB5132061 regression: RDP to Hyper-V VMs with GPU-P (AMD V710) disconnects ~1s (dxgkrnl 0xC000A003)

absalon vives 0 Reputation points
2026-11-14T07:37:24.58+00:00

Also filed via Feedback Hub (Hyper-V category).

Since cumulative update KB5132061, every RDP session to a Hyper-V guest that uses a GPU partition (GPU-P / WDDM GPU Partitioning) disconnects about 1 second after connecting. The Hyper-V console keeps working. Removing KB5132061 restores normal behaviour. Clear regression in dxgkrnl.

ENVIRONMENT

  • Host: Windows Server 2025 Datacenter, 24H2, build 26100.33438, Hyper-V.
  • Guests: Windows 11, 26H2, build 26300.9974 (build that includes KB5132061).
  • Shared GPU: AMD Radeon PRO V710 via GPU-P. Guest uses Microsoft Virtual Render Driver 10.0.26100.1150, paravirtualized to host AMD driver 32.0.21017.2052.

SYMPTOM

  • Each RDP session to a GPU-P guest disconnects ~1s after logon.
  • LiveKernelEvent 0x193; LKD_0x193_DxgkrnlLiveDump:802_Status_0xC000A003 ... failed_assign_resource_for_ID_device; session code 0x11.
  • Failure to assign the GPU resource to the display device of each RDP session (dxgkrnl).

REGRESSION ISOLATION (reproducible)

  • Remove KB5132061 + keep GPU-P -> OK
  • Keep KB5132061 + remove GPU-P -> OK
  • Keep KB5132061 + GPU-P active -> DISCONNECTS

=> break requires KB5132061 AND an active GPU partition. Root cause = dxgkrnl change in KB5132061.

RULED OUT

  • Driver: AMD package identical on host and in guest HostDriverStore (not a driver mismatch).
  • HVCI/VBS: disabling memory integrity makes the device go to OK (code 0) but RDP still dies -> not HVCI.
  • After removing KB5132061 guest reverts to build 26300.9711 and works again (GPU-P accelerates, Task Manager shows the V710 at real load/temperature).

REQUEST

  1. Is this KB5132061 x GPU-P regression known? Hotfix + ETA (before next cumulative)?
  2. Is GPU-P supported with a guest build newer than the host (Win11 26H2 guest on Server 2025 24H2 host)?
  3. Is the AMD Radeon PRO V710 in GPU-P officially supported on Server 2025?
  4. Any supported workaround other than removing the GPU partition?
Windows for business | Windows Server | Storage high availability | Virtualization and Hyper-V
0 comments No comments

1 answer

Sort by: Most helpful
  1. Daphne Huynh (WICLOUD CORPORATION) 2,620 Reputation points Microsoft External Staff Moderator
    2026-11-16T02:27:35.8966667+00:00

    Welcome to Microsoft Q&A!

    Thank you for taking the time to provide such detailed testing and isolation results.

    Your comparison between KB5132061 installed/removed and GPU-P enabled/removed is particularly helpful. Based on your results, there is a strong correlation between the updated guest build and an active GPU partition. However, there is an important distinction between demonstrating that correlation and confirming the underlying Windows code change as the official root cause.

    1. Is this a known KB5132061 × GPU-P regression? Is there a hotfix or ETA?

    At this time, I could not find this specific RDP + GPU-P + dxgkrnl 0xC000A003 issue documented as a known issue for KB5132061.

    Your tests provide strong evidence of an update-related interaction:

    • KB5132061 removed + GPU-P enabled → works
    • KB5132061 installed + GPU-P removed → works
    • KB5132061 installed + GPU-P enabled → RDP disconnects

    However, this is not the same as an officially confirmed root cause. The LiveKernelEvent and dxgkrnl dump would need to be analyzed by Microsoft Support and, if necessary, Windows engineering before a specific change in KB5132061 could be confirmed as the cause.

    Regarding a hotfix or ETA, Microsoft Q&A cannot provide or confirm one. Because Q&A is a public community support forum and does not have access to internal Windows bug records, engineering investigation status, unpublished fixes, or servicing schedules.

    I understand this may be frustrating given how thoroughly you have already isolated the behavior. Rather than speculate about an internal bug or give you an ETA that has not been published, I want to be clear about what can and cannot be confirmed from the information currently available.

    2. Is GPU-P supported when the Windows 11 guest is newer than the Windows Server host?

    Microsoft's published GPU-P requirements support Generation 2 VMs running Windows 10 or later, along with supported Windows Server guest operating systems.

    The published requirements do not state that the guest Windows release must be the same as, or older than, the Windows Server host release.

    Therefore, based on the published requirements, the fact that your Windows 11 26H2 guest is newer than the Windows Server 2025 host does not, by itself, establish that the GPU-P configuration is unsupported.

    3. Is AMD Radeon PRO V710 officially supported for GPU-P on Windows Server 2025?

    Yes. Microsoft's current GPU Partitioning documentation lists the AMD Radeon PRO V710 as a supported GPU for GPU-P.

    The hardware being supported does not remove the requirement to use the appropriate vendor-supported firmware and drivers, so the AMD driver package should also be validated for the applicable Windows Server GPU-P configuration.

    4. Is there a supported workaround other than removing GPU-P?

    At this time, I could not find a publicly documented Microsoft workaround that preserves this exact KB5132061 + RDP + AMD V710 GPU-P configuration while avoiding the disconnect.

    Since KB5132061 is a cumulative security update, I would not recommend permanently uninstalling it as the production solution.

    Based on the currently documented options and the results of your own testing, the conservative temporary workaround is therefore to keep the current Windows security update installed and temporarily remove the GPU partition from the affected VM while the issue is being investigated.

    As you have already performed much of the troubleshooting that would normally be recommended on the forum. Given your A/B testing, repeating generic RDP, HVCI, or driver troubleshooting would be unlikely to add much value.

    The remaining questions are different in nature: Has Microsoft internally confirmed the regression? What specific Windows code change caused it? Is a fix being developed? When will that fix be released?

    Those questions require access to Microsoft's internal engineering investigation and servicing information. Microsoft Q&A can help interpret published Microsoft documentation, validate documented support requirements, review troubleshooting evidence, and suggest the appropriate next step, but the community forum cannot access an internal Windows bug investigation or provide an unpublished hotfix status or release ETA.

    Your Feedback Hub submission is still useful because it provides product feedback and diagnostic information. However, if you need this issue formally investigated and tracked for a production environment, the appropriate next step is to open a Microsoft Support case.

    I recommend providing the Support engineer with your reproduction matrix, working and failing build numbers, AMD V710 driver versions, GPU-P configuration, relevant event logs, and the LiveKernelEvent 0x193 / 0xC000A003 / failed_assign_resource_for_ID_device dump information. This gives Support the evidence needed to determine whether the issue requires escalation to Windows engineering.

    To summarize, the currently published documentation supports GPU-P on Windows Server 2025, Windows 11 as a GPU-P guest, and AMD Radeon PRO V710 as a supported GPU-P device. Your testing also provides strong evidence of an interaction introduced with KB5132061. What Microsoft Q&A cannot confirm from public information is whether Microsoft Engineering has classified this specific behavior as a known regression, its exact dxgkrnl root cause, or whether a hotfix and release ETA exist.

    For that reason, opening a Microsoft Support case with the reproduction data and live dump is the one that I would like to recommend you for the next step.

    References: Troubleshooting guidance for Hyper-V GPU assignment, partitioning, and passthrough - Windows Server | Microsoft Learn

    Partition and share GPUs with virtual machines on Hyper-V | Microsoft Learn

    Plan for GPU acceleration in Windows Server | Microsoft Learn

    November 10, 2026—KB5132061 (OS Builds 26300.9974, 26200.9974 and 26100.9974) | Microsoft Support

    November 10, 2026—Hotpatch KB5132043 (OS Builds 26300.9911. 26200.9911, and 26100.9911) | Microsoft Support

    KB5132060: Safe OS Dynamic Update for Windows 11, versions 24H2, 25H2, and 26H2: November 10, 2026 | Microsoft Support

    Resolved issues in Windows 11, version 24H2 | Microsoft Learn

    Processor compatibility for Hyper-V virtual machines | Microsoft Learn

    If you find it useful, please click Accept Answer.

    Thank you for choosing Microsoft Q&A.

    Was this answer helpful?

    0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.