petri: add NVMe emulator support by babayet2 · Pull Request #3378 · microsoft/openvmm

babayet2 · 2026-04-26T22:36:49Z

Adds coverage for HyperV NVMe devices leveraging a closed source emulator

This test is currently marked as unstable, the CI runners are not guaranteed to have the NVMe emulator installed

Copilot

Pull request overview

Adds Hyper-V NVMe coverage in the Petri-based vmm_tests suite by provisioning NVMe devices via a closed-source Hyper-V “NVMe emulator” and wiring it through the Hyper-V Petri backend.

Changes:

Adds an (unstable) Hyper-V OpenHCL Linux storage test that validates discovery + IO via an NVMe emulator device.
Extends the Hyper-V Petri backend to translate VmbusStorageType::Nvme into post-create NVMe emulator device attachment.
Adds PowerShell-side plumbing to attach NVMe emulator devices after New-CustomVM completes.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 5 comments.

File	Description
`vmm_tests/vmm_tests/tests/tests/x86_64/storage.rs`	Adds an unstable Hyper-V NVMe emulator storage test that exercises discovery and IO.
`petri/src/vm/hyperv/mod.rs`	Maps `VmbusStorageType::Nvme` controllers to NVMe emulator device configs for Hyper-V backend runs.
`petri/src/vm/hyperv/powershell.rs`	Adds `nvme_emulator_devices` to VM creation args and attaches them post-create via PowerShell.

Copilot · 2026-04-26T22:41:38Z

+/// NVMe emulator device configuration (added via AddNvmeEmulator.ps1)
+pub struct HyperVNvmeEmulatorDevice {
+    /// Target VTL (0 or 2)
+    pub target_vtl: u8,
+    /// Emulator configuration strings (e.g., `["--vhd path1", "--vhd path2"]`)


HyperVNvmeEmulatorDevice is described as being added via AddNvmeEmulator.ps1, but the implementation below calls Import-Module HvlDeviceHost; Add-NvmeEmulator .... Please update the doc comment (and the nvme_emulator_devices field comment) to match the actual mechanism so future readers know where this device comes from and what dependency is required.

Suggested change

/// NVMe emulator device configuration (added via AddNvmeEmulator.ps1)

pub struct HyperVNvmeEmulatorDevice {

/// Target VTL (0 or 2)

pub target_vtl: u8,

/// Emulator configuration strings (e.g., `["--vhd path1", "--vhd path2"]`)

/// NVMe emulator device configuration for devices added via the

/// `Add-NvmeEmulator` PowerShell cmdlet from the `HvlDeviceHost` module.

pub struct HyperVNvmeEmulatorDevice {

/// Target VTL (0 or 2)

pub target_vtl: u8,

/// Emulator configuration strings passed to `Add-NvmeEmulator`

/// (for example, `["--vhd path1", "--vhd path2"]`).

///

/// Requires the `HvlDeviceHost` PowerShell module to be available so the

/// `Add-NvmeEmulator` cmdlet can be imported and invoked.

Copilot · 2026-04-26T22:41:39Z

+pub struct HyperVNvmeEmulatorDevice {
+    /// Target VTL (0 or 2)
+    pub target_vtl: u8,
+    /// Emulator configuration strings (e.g., `["--vhd path1", "--vhd path2"]`)
+    pub configuration_strings: Vec<String>,
+    /// VSID to assign to this device.
+    pub vsid: Guid,


HyperVNvmeEmulatorDevice.target_vtl is a u8 with an implicit constraint of "0 or 2". To avoid invalid values flowing through the API, consider modeling this as the existing crate::Vtl/petri::Vtl enum (and convert to the numeric form only at the PowerShell boundary), or introduce a small enum for the NVMe emulator target VTL.

Copilot · 2026-04-26T22:41:39Z

+    // (post-VM-creation). NVMe emulators cannot be added during DefineSystem
+    // and require a VSID workaround via ModifyResourceSettings.
+    for device in &args.nvme_emulator_devices {
+        run_add_nvme_emulator(&vmid, device).await?;


If run_add_nvme_emulator fails after New-CustomVM succeeds, run_new_customvm will return an error but the VM has already been created. This can leave stray VMs behind (especially in CI where the emulator/module may not be installed). Consider attempting best-effort cleanup on failure (e.g., ensure the VM is off and call Remove-VM) once you have a parsed vmid.

Suggested change

run_add_nvme_emulator(&vmid, device).await?;

if let Err(err) = run_add_nvme_emulator(&vmid, device).await {

let cleanup_err = run_remove_vm(&vmid).await.err();

return match cleanup_err {

Some(cleanup_err) => Err(err).context(format!(

"failed to add NVMe emulator device and failed to remove partially created VM {vmid}: {cleanup_err:#}"

)),

None => Err(err)

.context(format!("failed to add NVMe emulator device; removed partially created VM {vmid}")),

};

}

Copilot · 2026-04-26T22:41:39Z

+    let config_array = device
+        .configuration_strings
+        .iter()
+        .map(|s| format!("'{}'", s.replace('\'', "''")))
+        .collect::<Vec<_>>()
+        .join(", ");
+
+    let script = format!(
+        "Import-Module HvlDeviceHost; Add-NvmeEmulator -VmName '{}' -ConfigurationStrings @({}) -TargetVtl {} -Vsid '{{{}}}'",
+        vmid, config_array, device.target_vtl, device.vsid,
+    );
+
+    run_host_cmd(PowerShellBuilder::new().cmdlet(&script).finish().build())
+        .await
+        .map(|_| ())
+        .context("add_nvme_emulator")


run_add_nvme_emulator constructs a full PowerShell snippet via format!(...) and passes it as a raw cmdlet string. This bypasses the escaping/quoting guarantees provided by PowerShellBuilder/ps::Value (which this file already uses for arrays/hashtables), making it easier to introduce quoting bugs or injection issues if inputs ever contain unexpected characters. Prefer building the pipeline with PowerShellBuilder (including a ps::Array for ConfigurationStrings) and passing vmid/vsid/target_vtl as typed arguments.

Suggested change

let config_array = device

.configuration_strings

.iter()

.map(|s| format!("'{}'", s.replace('\'', "''")))

.collect::<Vec<_>>()

.join(", ");

let script = format!(

"Import-Module HvlDeviceHost; Add-NvmeEmulator -VmName '{}' -ConfigurationStrings @({}) -TargetVtl {} -Vsid '{{{}}}'",

vmid, config_array, device.target_vtl, device.vsid,

);

run_host_cmd(PowerShellBuilder::new().cmdlet(&script).finish().build())

.await

.map(|_| ())

.context("add_nvme_emulator")

let configuration_strings = ps::Array(

device

.configuration_strings

.iter()

.cloned()

.map(ps::Value::from)

.collect(),

);

run_host_cmd(

PowerShellBuilder::new()

.cmdlet("Import-Module")

.positional("HvlDeviceHost")

.next()

.cmdlet("Add-NvmeEmulator")

.arg("VmName", vmid.to_string())

.arg("ConfigurationStrings", configuration_strings)

.arg("TargetVtl", device.target_vtl)

.arg("Vsid", format!("{{{}}}", device.vsid))

.finish()

.build(),

)

.await

.map(|_| ())

.context("add_nvme_emulator")

Copilot · 2026-04-26T22:41:39Z

+                    // ordering — hvldevicehost assigns NSIDs 1..N by VHD
+                    // argument order.
+                    let mut sorted_drives: Vec<_> = drives.iter().collect();
+                    sorted_drives.sort_by_key(|(lun, _)| *lun);


In the VmbusStorageType::Nvme branch, the lun/key from drives is only used for sorting, but hvldevicehost assigns NSIDs sequentially (1..N) based on argument order. This means callers that specify explicit NSIDs (e.g., Some(5)) will not get the requested IDs, which can break the VTL2 backing device mapping that expects a specific NSID. Consider validating that the provided keys are exactly 1..N (or otherwise mapping/rewriting) and returning a clear error if the requested NSIDs can't be honored on Hyper-V.

Suggested change

sorted_drives.sort_by_key(|(lun, _)| *lun);

sorted_drives.sort_by_key(|(lun, _)| *lun);

// Hyper-V's NVMe emulator cannot honor arbitrary caller-

// supplied namespace IDs. It assigns NSIDs sequentially

// starting at 1 based on the VHD argument order, so only

// a requested mapping of 1..N can be represented.

for (index, (lun, _)) in sorted_drives.iter().enumerate() {

let expected_nsid = index + 1;

if **lun != expected_nsid {

anyhow::bail!(

"Hyper-V NVMe emulator requires namespace IDs to be exactly 1..N in argument order; requested NSID {} cannot be honored (expected {})",

lun,

expected_nsid

);

}

}

github-actions · 2026-04-26T23:46:52Z

3 Petri tests failed (3 unstable)

tjones60 · 2026-04-27T16:28:55Z

+    // (post-VM-creation). NVMe emulators cannot be added during DefineSystem
+    // and require a VSID workaround via ModifyResourceSettings.
+    for device in &args.nvme_emulator_devices {
+        run_add_nvme_emulator(&vmid, device).await?;


Please pass in the information needed to create the NVMe emulators as part of the original arguments to New-CustomVM so that we can add the appropriate ResourceSettings to the initial WMI DefineSystem call, rather than invoking a separate powershell session,

tjones60 · 2026-04-27T16:33:21Z

+        .join(", ");
+
+    let script = format!(
+        "Import-Module HvlDeviceHost; Add-NvmeEmulator -VmName '{}' -ConfigurationStrings @({}) -TargetVtl {} -Vsid '{{{}}}'",


For things like this that can easy be represented by the builder notation, please use that rather than the "script" backdoor.

Also, this pattern (of using a separate call to modify a nvme drive) should only be used as part of set_vmbus_drive, creation time call should be part of new-customvm for performance reasons.

Copilot

Pull request overview

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

Comments suppressed due to low confidence (1)

petri/src/vm/hyperv/mod.rs:389

ide_controllers is still computed above, but it is no longer passed into HyperVNewCustomVMArgs (only storage_controllers is set). This will silently drop IDE drive configuration for Hyper-V VMs and also leaves an unused ide_controllers local. Include the ide_controllers field in hyperv_args (or remove the mapping if IDE is intentionally unsupported).

            guest_state_path,
            storage_controllers,
            com_3: supports_com3,
            imc_hiv,
            management_vtl_settings,
            ..HyperVNewCustomVMArgs::from_config(&config, &properties)?
        };

        let vm = HyperVVM::new(hyperv_args, log_source.clone(), driver.clone()).await?;

github-actions · 2026-04-28T00:41:20Z

3 Petri tests failed (3 unstable)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

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

Comments suppressed due to low confidence (1)

petri/src/vm/hyperv/mod.rs:365

ide_controllers is built from the firmware config above, but it is no longer passed into HyperVNewCustomVMArgs (only storage_controllers is set here, and from_config initializes ide_controllers as empty). This silently drops IDE controller configuration for Hyper-V VMs and also leaves an unused local. Pass the computed ide_controllers into hyperv_args (or remove the mapping if it’s intentionally unsupported).

        let hyperv_args = HyperVNewCustomVMArgs {
            firmware_file: igvm_file.clone(),
            firmware_parameters: openhcl_command_line,
            guest_state_path,
            storage_controllers,
            com_3: supports_com3,
            imc_hiv,
            management_vtl_settings,
            ..HyperVNewCustomVMArgs::from_config(&config, &properties)?
        };

assert_eq! panics the test runner instead of returning a structured error. Use anyhow::ensure! so NSID validation failures surface as clean error messages. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

github-actions · 2026-04-28T17:59:16Z

3 Petri tests failed (3 unstable)

babayet2 assigned tjones60 Apr 26, 2026

Copilot AI review requested due to automatic review settings April 26, 2026 22:36

babayet2 requested a review from a team as a code owner April 26, 2026 22:36

Copilot started reviewing on behalf of babayet2 April 26, 2026 22:37 View session

Copilot AI reviewed Apr 26, 2026

View reviewed changes

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/mod.rs Outdated

babayet2 force-pushed the emulated-nvme branch from 608577d to a39fb2d Compare April 27, 2026 20:57

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/mod.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/mod.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/hyperv.psm1 Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/mod.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/hyperv.psm1

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

Copilot AI review requested due to automatic review settings April 27, 2026 23:31

babayet2 force-pushed the emulated-nvme branch from a39fb2d to 87af4b4 Compare April 27, 2026 23:31

Copilot started reviewing on behalf of babayet2 April 27, 2026 23:31 View session

Copilot AI reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

Comment thread petri/src/vm/hyperv/hyperv.psm1 Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

tjones60 reviewed Apr 27, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/mod.rs Outdated

tjones60 reviewed Apr 28, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

tjones60 reviewed Apr 28, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/mod.rs Outdated

tjones60 reviewed Apr 28, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

babayet2 force-pushed the emulated-nvme branch from 1fb3196 to d4605ff Compare April 28, 2026 16:07

petri: add NVMe emulator support for Hyper-V

a176438

babayet2 force-pushed the emulated-nvme branch from d4605ff to a176438 Compare April 28, 2026 16:09

babayet2 and others added 2 commits April 28, 2026 16:17

fmt fix

9d7e87f

fix(petri): check HvlDeviceHost module availability before import

2fa8b23

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings April 28, 2026 16:35

Copilot started reviewing on behalf of babayet2 April 28, 2026 16:36 View session

Copilot AI reviewed Apr 28, 2026

View reviewed changes

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

Comment thread petri/src/vm/hyperv/powershell.rs

Comment thread petri/src/vm/hyperv/powershell.rs Outdated

-        run_add_nvme_emulator(&vmid, device).await?;
+        if let Err(err) = run_add_nvme_emulator(&vmid, device).await {
+            let cleanup_err = run_remove_vm(&vmid).await.err();
+            return match cleanup_err {
+                Some(cleanup_err) => Err(err).context(format!(
+                    "failed to add NVMe emulator device and failed to remove partially created VM {vmid}: {cleanup_err:#}"
+                )),
+                None => Err(err)
+                    .context(format!("failed to add NVMe emulator device; removed partially created VM {vmid}")),
+            };
+        }

-                    sorted_drives.sort_by_key(|(lun, _)| *lun);
+                    sorted_drives.sort_by_key(|(lun, _)| *lun);
+                    // Hyper-V's NVMe emulator cannot honor arbitrary caller-
+                    // supplied namespace IDs. It assigns NSIDs sequentially
+                    // starting at 1 based on the VHD argument order, so only
+                    // a requested mapping of 1..N can be represented.
+                    for (index, (lun, _)) in sorted_drives.iter().enumerate() {
+                        let expected_nsid = index + 1;
+                        if **lun != expected_nsid {
+                            anyhow::bail!(
+                                "Hyper-V NVMe emulator requires namespace IDs to be exactly 1..N in argument order; requested NSID {} cannot be honored (expected {})",
+                                lun,
+                                expected_nsid
+                            );
+                        }
+                    }

Conversation

babayet2 commented Apr 26, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Apr 26, 2026

Uh oh!

tjones60 Apr 27, 2026

Choose a reason for hiding this comment

Uh oh!

tjones60 Apr 27, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

github-actions Bot commented Apr 28, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

github-actions Bot commented Apr 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants