Implement MSHV aarch64 backend by cshung · Pull Request #1626 · hyperlight-dev/hyperlight

cshung · 2026-07-06T22:39:28Z

Summary

Implements the VirtualMachine trait for MSHV on aarch64, replacing the previous stub.

Changes

mshv/aarch64.rs: Full implementation — VM creation, memory mapping, vcpu run loop with MMIO-based I/O dispatch, register access (general/FPU/special via hv_register_assoc), vcpu reset
hyperlight_vm/aarch64.rs: Wire up MshvVm::new() when MSHV is detected
memory_region.rs: Implement TryFrom<hv_arm64_memory_intercept_message> for MemoryRegionFlags

Design

Follows KVM aarch64 I/O page pattern (MMIO writes to I/O page GPA → port I/O)
Follows MSHV x86_64 patterns for partition creation and register access
Uses low-level get_reg/set_reg for system regs (get_sregs/set_sregs not available on aarch64 in mshv-ioctls)

Testing

Cross-compiled and checked for aarch64-unknown-linux-gnu locally ✅
No MSHV aarch64 runtime environment available yet for integration testing

Add a full VirtualMachine trait implementation for MSHV on aarch64, replacing the previous stub that returned false/unimplemented. Key changes: - mshv/aarch64.rs: Complete implementation including VM creation, memory mapping, vcpu run loop with MMIO-based I/O dispatch, general/FPU/special register access via hv_register_assoc, and vcpu reset support. - hyperlight_vm/aarch64.rs: Wire up MshvVm::new() so MSHV is used when detected (previously returned NoHypervisorFound). - memory_region.rs: Implement TryFrom<hv_arm64_memory_intercept_message> for MemoryRegionFlags using intercept_access_type field. The implementation mirrors the KVM aarch64 backend's I/O page pattern (MMIO writes to the I/O page GPA are interpreted as port I/O) and follows MSHV x86_64 patterns for partition creation and register access. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

Configure the GIC, decode MMIO writes, preserve register state, and rebuild the partition during restore. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 6f20a05d-6bee-4e2e-b320-12f8d9759bbc Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

hyperlight-gh-bot · 2026-11-02T17:53:22Z

Benchmark Results

Measured commit: 2f42d322fcc6
Baseline commit: fa13116cd998

kvm / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`decode_vec_bytes_copy`	`encode_control`
	`373.22 µs` (➖ 1.00x faster)
`byte_chunks`		`774.35 ns` (➖ 1.01x slower)
`vec_bytes`		`581.30 ns` (➖ 1.03x faster)

payload_allocation

	`slot_pool_segmented`
`65536`	`143.53 ns` (➖ 1.01x slower)
`262144`	`525.28 ns` (➖ 1.01x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`13.36 ms` (➖ 1.01x slower)

slot_pool

	`alloc_dealloc_4096`	`alloc_dealloc_128`	`alloc_dealloc_1500`
	`7.83 ns` (➖ 1.01x slower)	`7.68 ns` (➖ 1.03x faster)	`7.68 ns` (➖ 1.00x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`100.85 µs` (➖ 1.00x slower)

virtq_readonly

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`65536`	`1.95 µs` (➖ 1.06x faster)	`2.03 µs` (➖ 1.06x slower)
`262144`	`7.20 µs` (➖ 1.02x faster)	`7.25 µs` (➖ 1.00x slower)

virtq_readwrite

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`8192`	`1.08 µs` (➖ 1.02x slower)	`1.05 µs` (➖ 1.05x faster)
`65536`	`6.19 µs` (➖ 1.04x faster)	`6.27 µs` (➖ 1.04x faster)
`262144`	`26.90 µs` (➖ 1.01x faster)

kvm / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`decode_vec_bytes_copy`	`encode_control`
	`762.84 µs` (➖ 1.01x slower)
`byte_chunks`		`791.78 ns` (➖ 1.04x faster)
`vec_bytes`		`612.47 ns` (➖ 1.01x slower)

payload_allocation

	`slot_pool_segmented`
`65536`	`204.75 ns` (➖ 1.00x slower)
`262144`	`594.41 ns` (➖ 1.00x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`15.60 ms` (➖ 1.00x faster)

slot_pool

	`alloc_dealloc_4096`	`alloc_dealloc_128`	`alloc_dealloc_1500`
	`8.10 ns` (➖ 1.01x faster)	`8.13 ns` (➖ 1.00x faster)	`8.10 ns` (➖ 1.01x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`56.41 µs` (➖ 1.05x faster)

virtq_readonly

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`65536`	`2.54 µs` (➖ 1.01x faster)	`2.53 µs` (➖ 1.01x faster)
`262144`	`8.99 µs` (➖ 1.00x faster)	`8.97 µs` (➖ 1.01x faster)

virtq_readwrite

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`8192`	`842.41 ns` (➖ 1.01x slower)	`853.13 ns` (➖ 1.04x faster)
`65536`	`8.37 µs` (➖ 1.02x faster)	`8.39 µs` (➖ 1.02x faster)
`262144`	`34.83 µs` (➖ 1.00x faster)

mshv3 / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`decode_vec_bytes_copy`	`encode_control`
	`314.96 µs` (➖ 1.19x faster)
`byte_chunks`		`928.72 ns` (➖ 1.05x slower)
`vec_bytes`		`667.87 ns` (➖ 1.05x faster)

payload_allocation

	`slot_pool_segmented`
`65536`	`182.56 ns` (➖ 1.00x faster)
`262144`	`670.67 ns` (➖ 1.00x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`63.16 ms` (➖ 1.12x faster)

slot_pool

	`alloc_dealloc_4096`	`alloc_dealloc_128`	`alloc_dealloc_1500`
	`9.87 ns` (➖ 1.00x faster)	`9.88 ns` (➖ 1.00x faster)	`9.89 ns` (➖ 1.00x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`109.23 µs` (➖ 1.04x faster)

virtq_readonly

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`65536`	`2.48 µs` (➖ 1.00x slower)	`2.63 µs` (➖ 1.02x faster)
`262144`	`9.03 µs` (➖ 1.06x faster)	`12.02 µs` (➖ 1.28x slower)

virtq_readwrite

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`8192`	`1.32 µs` (➖ 1.04x slower)	`1.38 µs` (➖ 1.06x slower)
`65536`	`7.93 µs` (➖ 1.02x faster)	`7.98 µs` (➖ 1.01x faster)
`262144`	`34.92 µs` (➖ 1.05x faster)

mshv3 / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`decode_vec_bytes_copy`	`encode_control`
	`660.67 µs` (➖ 1.03x slower)
`byte_chunks`		`923.31 ns` (➖ 1.02x slower)
`vec_bytes`		`688.79 ns` (➖ 1.04x slower)

payload_allocation

	`slot_pool_segmented`
`65536`	`260.13 ns` (➖ 1.00x faster)
`262144`	`739.38 ns` (➖ 1.10x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`64.64 ms` (➖ 1.20x faster)

slot_pool

	`alloc_dealloc_4096`	`alloc_dealloc_128`	`alloc_dealloc_1500`
	`8.67 ns` (➖ 1.00x faster)	`8.69 ns` (➖ 1.01x faster)	`8.68 ns` (➖ 1.00x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`53.62 µs` (➖ 1.06x faster)

virtq_readonly

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`65536`	`2.25 µs` (➖ 1.01x faster)	`2.28 µs` (➖ 1.00x slower)
`262144`	`7.92 µs` (➖ 1.00x faster)	`7.89 µs` (➖ 1.02x faster)

virtq_readwrite

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`8192`	`894.66 ns` (➖ 1.01x faster)	`903.16 ns` (➖ 1.02x faster)
`65536`	`7.43 µs` (➖ 1.02x faster)	`7.54 µs` (➖ 1.00x faster)
`262144`	`36.67 µs` (➖ 1.05x faster)

hyperv-ws2025 / amd (Windows) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`decode_vec_bytes_copy`	`encode_control`
	`1.96 ms` (➖ 1.01x slower)
`byte_chunks`		`1.21 µs` (➖ 1.01x faster)
`vec_bytes`		`821.72 ns` (➖ 1.05x slower)

payload_allocation

	`slot_pool_segmented`
`65536`	`270.93 ns` (➖ 1.18x slower)
`262144`	`814.11 ns` (➖ 1.05x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`12.86 ms` (➖ 1.05x slower)

slot_pool

	`alloc_dealloc_4096`	`alloc_dealloc_128`	`alloc_dealloc_1500`
	`10.00 ns` (➖ 1.00x slower)	`10.15 ns` (➖ 1.02x slower)	`10.14 ns` (➖ 1.01x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`853.51 µs` (➖ 1.07x slower)

virtq_readonly

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`65536`	`2.36 µs` (➖ 1.08x faster)	`2.36 µs` (➖ 1.00x faster)
`262144`	`9.28 µs` (➖ 1.07x faster)	`9.47 µs` (➖ 1.08x faster)

virtq_readwrite

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`8192`	`1.30 µs` (➖ 1.02x slower)	`1.30 µs` (➖ 1.06x slower)
`65536`	`8.95 µs` (➖ 1.05x faster)	`9.13 µs` (➖ 1.03x faster)
`262144`	`44.52 µs` (➖ 1.05x slower)

hyperv-ws2025 / intel (Windows) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`decode_vec_bytes_copy`	`encode_control`
	`2.87 ms` (➖ 1.03x faster)
`byte_chunks`		`1.16 µs` (➖ 1.00x faster)
`vec_bytes`		`754.06 ns` (➖ 1.01x faster)

payload_allocation

	`slot_pool_segmented`
`65536`	`214.16 ns` (➖ 1.03x slower)
`262144`	`743.79 ns` (➖ 1.06x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`13.33 ms` (➖ 1.07x faster)

slot_pool

	`alloc_dealloc_4096`	`alloc_dealloc_128`	`alloc_dealloc_1500`
	`10.30 ns` (➖ 1.01x slower)	`9.70 ns` (➖ 1.04x faster)	`9.75 ns` (➖ 1.02x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`607.48 µs` (➖ 1.09x faster)

virtq_readonly

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`65536`	`2.63 µs` (➖ 1.13x slower)	`2.31 µs` (➖ 1.00x slower)
`262144`	`7.61 µs` (➖ 1.00x faster)	`7.70 µs` (➖ 1.02x slower)

virtq_readwrite

	`slot_pool_segmented`	`slot_pool_segmented_fragmented`
`8192`	`1.10 µs` (➖ 1.04x slower)	`1.19 µs` (➖ 1.10x slower)
`65536`	`8.97 µs` (➖ 1.06x slower)	`8.79 µs` (➖ 1.05x slower)
`262144`	`44.44 µs` (➖ 1.10x faster)

_{Reported by cargo ci bench-report --candidate run:41488633866 --config-file bench_report.toml.}

cshung added the kind/enhancement For PRs adding features, improving functionality, docs, tests, etc. label Jul 7, 2026

cshung force-pushed the cshung/mshv-aarch64 branch from b34cc5d to fb92ede Compare July 31, 2026 16:13

cshung force-pushed the cshung/mshv-aarch64 branch from fb92ede to 3736e16 Compare August 12, 2026 14:55

cshung force-pushed the cshung/mshv-aarch64 branch 2 times, most recently from b440e24 to 4ada969 Compare August 20, 2026 17:42

cshung force-pushed the cshung/mshv-aarch64 branch from 4ada969 to 7f2289d Compare September 2, 2026 22:47

cshung force-pushed the cshung/mshv-aarch64 branch from a70dcfa to 1b275f2 Compare September 29, 2026 05:30

cshung force-pushed the cshung/mshv-aarch64 branch from 1b275f2 to 2f42d32 Compare November 2, 2026 17:10

cshung and others added 7 commits November 2, 2026 09:40

fix: harden ARM64 MSHV backend

387aa6d

Configure the GIC, decode MMIO writes, preserve register state, and rebuild the partition during restore. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

Match mutable virtual machine register setters

aeaf732

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 6f20a05d-6bee-4e2e-b320-12f8d9759bbc Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

ci: probe WSL on Windows ARM64

d0bd5dd

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 6f20a05d-6bee-4e2e-b320-12f8d9759bbc Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

ci: attempt WSL installation

dd645fd

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 6f20a05d-6bee-4e2e-b320-12f8d9759bbc Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

ci: enable WSL component

d5f6be9

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 6f20a05d-6bee-4e2e-b320-12f8d9759bbc Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

Revert temporary ARM64 WSL probe

2f42d32

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 6f20a05d-6bee-4e2e-b320-12f8d9759bbc Signed-off-by: cshung <3410332+cshung@users.noreply.github.com>

Conversation

cshung commented Jul 6, 2026

Summary

Changes

Design

Testing

Uh oh!

hyperlight-gh-bot Bot commented Nov 2, 2026

Benchmark Results

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Benchmark Results

function_call_codec

payload_allocation

sandboxes

slot_pool

snapshot_files

virtq_readonly

virtq_readwrite

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant