Run guard page crash tests on Windows by DanielTangnes · Pull Request #1899 · hyperlight-dev/hyperlight

DanielTangnes · 2026-10-07T21:04:51Z

guard_page_crash_test (read/write/exec on the leading guard page of ExclusiveSharedMemory) runs on Linux and Windows. It uses the subprocess approach from #1498.

What changed:

test_path, run_guard_page_subprocess and killed_by_access_violation live in a small guard_page_subprocess test module that both shims use. The bodies are unchanged, so git diff --color-moved shows them as a move.
The three tests print reached_guard / survived_guard around the bad access. The shim reruns the current test binary with --ignored --exact for each one.
The module is gated on all(not(miri), any(target_os = "linux", windows)).
Removed the signal-hook-registry dev-dependency. The SIGSEGV hook in these tests was its only user.

A side effect on Linux: the old shim spawned cargo test without --profile, so under just test release it rebuilt and checked a debug binary. Rerunning current_exe() keeps the child on the parent's profile and skips the extra build.

I left macOS out to match the issue. from_file_tests already runs there, but I couldn't check which signal the exec case raises on Apple silicon. Happy to widen the gate if you want it.

Test-only change, so no changelog entry.

Testing (Linux x86_64, KVM host):

cargo test -p hyperlight-host --lib -- mem::shared_mem::tests: 28 passed, 5 ignored
cargo test -p hyperlight-host --profile=release --lib -- guard_page: both shims pass
Sanity check: with the leading guard page left PROT_READ | PROT_WRITE, the shim fails with survived=true (reverted afterwards)
rustfmt and clippy: no findings in shared_mem.rs. I ran these with a 1.98 toolchain rather than the pinned 1.94 and nightly rustfmt, so CI is the real check.
Not run locally: Windows, macOS, just miri-tests, full just test. The Windows run comes from CI.

Could a maintainer add kind/enhancement? I can't set labels from a fork.

🤖 Generated with Claude Code

Copilot

🟢 Approval recommended

The subprocess checks, platform gates, and dependency cleanup are consistent and complete.

0 open findings

What changed in this PR

Enables ExclusiveSharedMemory guard page crash tests on Windows using isolated test subprocesses.

Changes:

Reuses a shared subprocess test harness.
Detects Linux signals and Windows access violations.
Removes the unused direct signal hook dependency.

File	Description
`src/hyperlight_host/src/mem/shared_mem.rs`	Adds cross-platform guard page crash testing.
`src/hyperlight_host/Cargo.toml`	Removes the unused dev dependency.
`Cargo.lock`	Removes the direct dependency edge.

🧠 Review effort: Balanced

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

ludfjig

thanks a lot for this, this is great!

hyperlight-gh-bot · 2026-10-12T18:07:54Z

Benchmark Results

Measured commit: edca0d5cb1c6
Baseline commit: 5ddb37af894e

kvm / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`791.84 ns` (➖ 1.04x slower)
`vec_bytes`	`586.78 ns` (➖ 1.00x slower)
		`375.35 µs` (➖ 1.00x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`524.65 ns` (➖ 1.07x faster)
`65536`	`141.62 ns` (➖ 1.02x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`78.58 ms` (➖ 1.01x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`7.74 ns` (➖ 1.00x faster)	`7.75 ns` (➖ 1.00x faster)	`7.89 ns` (➖ 1.06x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`92.10 µs` (➖ 1.02x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.20 µs` (➖ 1.01x faster)	`7.16 µs` (➖ 1.05x faster)
`65536`	`2.06 µs` (➖ 1.02x faster)	`1.94 µs` (➖ 1.00x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`6.29 µs` (➖ 1.01x slower)	`6.25 µs` (➖ 1.02x faster)
`8192`	`1.06 µs` (➖ 1.03x slower)	`1.06 µs` (➖ 1.01x faster)
`262144`		`27.17 µs` (➖ 1.00x slower)

kvm / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`826.45 ns` (➖ 1.18x slower)
`vec_bytes`	`607.69 ns` (➖ 1.17x slower)
		`762.40 µs` (➖ 1.17x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`585.31 ns` (➖ 1.16x slower)
`65536`	`157.21 ns` (➖ 1.14x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`81.59 ms` (➖ 1.14x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`8.12 ns` (➖ 1.17x slower)	`8.00 ns` (➖ 1.16x slower)	`8.09 ns` (➖ 1.16x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`53.13 µs` (➖ 1.20x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`8.63 µs` (➖ 1.14x slower)	`8.72 µs` (➖ 1.16x slower)
`65536`	`2.44 µs` (➖ 1.16x slower)	`2.54 µs` (➖ 1.17x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.09 µs` (➖ 1.15x slower)	`8.32 µs` (➖ 1.17x slower)
`8192`	`846.98 ns` (➖ 1.18x slower)	`856.57 ns` (➖ 1.18x slower)
`262144`		`33.40 µs` (➖ 1.14x slower)

mshv3 / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`951.72 ns` (➖ 1.00x slower)
`vec_bytes`	`715.24 ns` (➖ 1.00x slower)
		`294.60 µs` (➖ 1.24x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`734.61 ns` (➖ 1.01x slower)
`65536`	`192.25 ns` (➖ 1.01x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`62.27 ms` (➖ 1.09x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.74 ns` (➖ 1.00x slower)	`9.73 ns` (➖ 1.00x slower)	`9.68 ns` (➖ 1.00x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`88.21 µs` (➖ 1.06x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`9.04 µs` (➖ 1.06x faster)	`8.89 µs` (➖ 1.09x faster)
`65536`	`2.25 µs` (➖ 1.07x faster)	`2.24 µs` (➖ 1.09x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.18 µs` (➖ 1.05x faster)	`8.43 µs` (➖ 1.05x slower)
`8192`	`1.27 µs` (➖ 1.02x faster)	`1.32 µs` (➖ 1.02x faster)
`262144`		`36.52 µs` (➖ 1.01x faster)

mshv3 / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`930.94 ns` (➖ 1.00x faster)
`vec_bytes`	`677.75 ns` (➖ 1.01x slower)
		`695.70 µs` (➖ 1.00x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`621.56 ns` (➖ 1.10x faster)
`65536`	`169.98 ns` (➖ 1.00x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`64.20 ms` (➖ 1.00x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.79 ns` (➖ 1.00x faster)	`8.76 ns` (➖ 1.00x faster)	`8.78 ns` (➖ 1.00x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`43.96 µs` (➖ 1.03x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.80 µs` (➖ 1.01x faster)	`7.75 µs` (➖ 1.00x faster)
`65536`	`2.25 µs` (➖ 1.01x slower)	`2.27 µs` (➖ 1.01x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`7.56 µs` (➖ 1.00x slower)	`7.56 µs` (➖ 1.01x faster)
`8192`	`857.88 ns` (➖ 1.00x faster)	`865.31 ns` (➖ 1.03x faster)
`262144`		`38.18 µs` (➖ 1.07x slower)

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

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`1.36 µs` (➖ 1.30x slower)
`vec_bytes`	`756.70 ns` (➖ 1.22x faster)
		`2.38 ms` (➖ 1.08x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`784.26 ns` (➖ 1.01x faster)
`65536`	`230.51 ns` (➖ 1.00x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`108.92 ms` (➖ 1.28x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`12.46 ns` (➖ 1.14x slower)	`11.37 ns` (➖ 1.07x slower)	`10.57 ns` (➖ 1.06x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`780.12 µs` (➖ 1.14x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`8.68 µs` (➖ 1.08x faster)	`8.69 µs` (➖ 1.24x faster)
`65536`	`2.54 µs` (➖ 1.05x slower)	`2.56 µs` (➖ 1.08x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.93 µs` (➖ 1.08x slower)	`8.98 µs` (➖ 1.01x slower)
`8192`	`1.43 µs` (➖ 1.05x slower)	`1.36 µs` (➖ 1.05x slower)
`262144`		`46.21 µs` (➖ 1.08x slower)

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

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`1.22 µs` (➖ 1.18x slower)
`vec_bytes`	`836.70 ns` (➖ 1.10x slower)
		`3.81 ms` (➖ 1.21x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`742.87 ns` (➖ 1.38x faster)
`65536`	`211.29 ns` (➖ 1.45x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`108.85 ms` (➖ 1.00x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`10.05 ns` (➖ 1.31x faster)	`10.62 ns` (➖ 1.38x faster)	`10.03 ns` (➖ 1.28x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`621.14 µs` (➖ 1.07x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.79 µs` (➖ 1.02x slower)	`7.83 µs` (➖ 1.03x faster)
`65536`	`2.33 µs` (➖ 1.02x slower)	`2.37 µs` (➖ 1.01x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.72 µs` (➖ 1.00x slower)	`9.29 µs` (➖ 1.07x slower)
`8192`	`1.14 µs` (➖ 1.09x faster)	`1.22 µs` (➖ 1.08x slower)
`262144`		`59.56 µs` (➖ 1.42x slower)

_{Reported by cargo ci bench-report --candidate run:37686822795 --baseline run:37707245718 --config-file bench_report.toml.}

ludfjig · 2026-10-12T18:08:50Z

@DanielTangnes looks like CI is failing. If you could address the formatting and then rebase I'd be happy to merge this :)

The read, write and exec guard page tests rerun the test binary with --exact and check its exit status using the helpers from the from_file guard tests. That works on Windows, where the SIGSEGV hook can't, and keeps the child on the same build profile as the parent. Drop the signal-hook-registry dev-dependency. These tests were its only user. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011CnFoNrC4zWqRUqoeDzS6P Signed-off-by: danieltangnes <danieltangnes1@gmail.com>

DanielTangnes · 2026-10-14T07:40:54Z

@ludfjig Should be good to go now :)

DanielTangnes marked this pull request as ready for review October 7, 2026 21:11

Copilot AI balanced review requested due to automatic review settings October 7, 2026 21:11

DanielTangnes requested review from andreiltd, danbugs, dblnz, devigned, jprendes, jsturtevant, ludfjig, simongdavies, squillace and syntactically as code owners October 7, 2026 21:11

Copilot started reviewing on behalf of DanielTangnes October 7, 2026 21:12 View session

Copilot AI reviewed Oct 7, 2026

View reviewed changes

ludfjig added the kind/enhancement For PRs adding features, improving functionality, docs, tests, etc. label Oct 12, 2026

ludfjig previously approved these changes Oct 12, 2026

View reviewed changes

DanielTangnes dismissed ludfjig’s stale review via 7bd0b1a October 14, 2026 07:25

DanielTangnes force-pushed the windows-guard-page-tests branch from edca0d5 to 7bd0b1a Compare October 14, 2026 07:25

Conversation

DanielTangnes commented Oct 7, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟢 Approval recommended

Uh oh!

ludfjig left a comment

Choose a reason for hiding this comment

Uh oh!

hyperlight-gh-bot Bot commented Oct 12, 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!

ludfjig commented Oct 12, 2026

Uh oh!

DanielTangnes commented Oct 14, 2026

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants