vmem: Add stronger tlb invalidation barrier by syntactically · Pull Request #1876 · hyperlight-dev/hyperlight

syntactically · 2026-10-02T10:52:16Z

No description provided.

Copilot

Copilot review overview

🟡 Changes recommended

Unaligned ranges can leave stale permissions cached, and the new architecture-specific behavior lacks tests.

Review effort: Balanced
Findings: 1 · 1 · 2

Open (4)

Require or normalize page-aligned invalidation bounds · New
Add cross-architecture integration tests for invalidation · New
Add SAFETY comment for EL1 maintenance assembly · New
Add SAFETY comment for privileged assembly block · New

What changed in this PR

Adds a guest paging barrier for invalidating cached translations after permission downgrades.

Changes:

Exposes downgrade_in_place.
Uses invlpg on amd64.
Uses DSB, TLBI, and ISB on AArch64.

File	Description
`src/hyperlight_guest_bin/src/paging.rs`	Exposes the barrier API.
`src/hyperlight_guest_bin/src/arch/amd64/paging.rs`	Adds amd64 invalidation.
`src/hyperlight_guest_bin/src/arch/aarch64/paging.rs`	Adds AArch64 invalidation.

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

Signed-off-by: Lucy Menon <168595099+syntactically@users.noreply.github.com>

hyperlight-gh-bot · 2026-10-02T16:16:17Z

Benchmark Results

Measured commit: 3423312ebafc
Baseline commit: 43cf3539bd10

kvm / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`766.32 ns` (➖ 1.01x faster)
`vec_bytes`	`587.62 ns` (➖ 1.02x slower)
		`375.62 µs` (➖ 1.00x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`543.74 ns` (➖ 1.02x slower)
`65536`	`141.37 ns` (➖ 1.02x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`77.52 ms` (➖ 1.03x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`7.85 ns` (➖ 1.00x slower)	`7.73 ns` (➖ 1.01x slower)	`7.78 ns` (➖ 1.00x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`93.76 µs` (➖ 1.01x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.26 µs` (➖ 1.01x slower)	`7.31 µs` (➖ 1.00x faster)
`65536`	`2.07 µs` (➖ 1.00x slower)	`2.08 µs` (➖ 1.08x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`6.32 µs` (➖ 1.01x faster)	`6.25 µs` (➖ 1.02x faster)
`8192`	`1.06 µs` (➖ 1.00x faster)	`1.08 µs` (➖ 1.01x slower)
`262144`		`27.43 µs` (➖ 1.01x slower)

kvm / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`683.51 ns` (➖ 1.09x faster)
`vec_bytes`	`518.06 ns` (➖ 1.05x faster)
		`691.40 µs` (➖ 1.00x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`515.66 ns` (➖ 1.11x faster)
`65536`	`139.34 ns` (➖ 1.11x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`76.25 ms` (➖ 1.04x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`7.06 ns` (➖ 1.06x faster)	`7.05 ns` (➖ 1.06x faster)	`7.06 ns` (➖ 1.06x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`46.96 µs` (➖ 1.05x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.76 µs` (➖ 1.10x faster)	`7.81 µs` (➖ 1.10x faster)
`65536`	`2.18 µs` (➖ 1.10x faster)	`2.20 µs` (➖ 1.08x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`7.31 µs` (➖ 1.08x faster)	`7.17 µs` (➖ 1.11x faster)
`8192`	`760.88 ns` (➖ 1.09x faster)	`737.95 ns` (➖ 1.08x faster)
`262144`		`30.11 µs` (➖ 1.11x faster)

mshv3 / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`958.90 ns` (➖ 1.01x slower)
`vec_bytes`	`712.22 ns` (➖ 1.00x faster)
		`336.21 µs` (➖ 1.05x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`727.24 ns` (➖ 1.01x faster)
`65536`	`194.19 ns` (➖ 1.01x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`58.16 ms` (➖ 1.05x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.77 ns` (➖ 1.02x slower)	`9.74 ns` (➖ 1.00x slower)	`9.72 ns` (➖ 1.06x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`83.84 µs` (➖ 1.00x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`9.05 µs` (➖ 1.07x faster)	`8.97 µs` (➖ 1.01x faster)
`65536`	`2.29 µs` (➖ 1.01x slower)	`2.46 µs` (➖ 1.00x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.24 µs` (➖ 1.00x slower)	`8.70 µs` (➖ 1.08x slower)
`8192`	`1.35 µs` (➖ 1.06x slower)	`1.34 µs` (➖ 1.01x slower)
`262144`		`36.64 µs` (➖ 1.00x slower)

mshv3 / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`926.64 ns` (➖ 1.00x faster)
`vec_bytes`	`653.34 ns` (➖ 1.01x faster)
		`685.95 µs` (➖ 1.02x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`626.43 ns` (➖ 1.03x faster)
`65536`	`165.46 ns` (➖ 1.01x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`66.12 ms` (➖ 1.06x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`8.75 ns` (➖ 1.00x faster)	`8.75 ns` (➖ 1.00x faster)	`8.76 ns` (➖ 1.00x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`43.22 µs` (➖ 1.02x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.74 µs` (➖ 1.00x faster)	`7.97 µs` (➖ 1.00x faster)
`65536`	`2.21 µs` (➖ 1.01x faster)	`2.27 µs` (➖ 1.00x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`7.71 µs` (➖ 1.03x slower)	`7.55 µs` (➖ 1.00x faster)
`8192`	`853.74 ns` (➖ 1.00x faster)	`880.94 ns` (➖ 1.01x faster)
`262144`		`36.91 µs` (➖ 1.03x faster)

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

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`1.21 µs` (➖ 1.07x slower)
`vec_bytes`	`820.84 ns` (➖ 1.15x faster)
		`2.19 ms` (➖ 1.14x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`828.97 ns` (➖ 1.02x slower)
`65536`	`231.90 ns` (➖ 1.02x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`87.66 ms` (➖ 1.26x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`10.04 ns` (➖ 1.38x faster)	`10.12 ns` (➖ 1.04x faster)	`10.12 ns` (➖ 1.04x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`714.64 µs` (➖ 1.11x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`9.13 µs` (➖ 1.05x slower)	`9.16 µs` (➖ 1.02x faster)
`65536`	`2.38 µs` (➖ 1.06x faster)	`2.32 µs` (➖ 1.16x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.82 µs` (➖ 1.09x slower)	`8.75 µs` (➖ 1.00x slower)
`8192`	`1.26 µs` (➖ 1.13x faster)	`1.28 µs` (➖ 1.10x faster)
`262144`		`38.89 µs` (➖ 1.09x slower)

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

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`1.21 µs` (➖ 1.00x slower)
`vec_bytes`	`800.22 ns` (➖ 1.04x slower)
		`3.14 ms` (➖ 1.01x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`745.59 ns` (➖ 1.01x slower)
`65536`	`213.97 ns` (➖ 1.01x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`108.79 ms` (➖ 1.11x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`10.01 ns` (➖ 1.00x slower)	`10.20 ns` (➖ 1.01x slower)	`10.66 ns` (➖ 1.00x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`615.62 µs` (➖ 1.01x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.79 µs` (➖ 1.02x slower)	`7.71 µs` (➖ 1.01x faster)
`65536`	`2.31 µs` (➖ 1.03x slower)	`2.32 µs` (➖ 1.00x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.55 µs` (➖ 1.02x slower)	`10.00 µs` (➖ 1.18x slower)
`8192`	`1.16 µs` (➖ 1.08x faster)	`1.21 µs` (➖ 1.08x slower)
`262144`		`52.15 µs` (➖ 1.22x slower)

_{Reported by cargo ci bench-report --candidate run:37029632479 --baseline run:36945564806 --config-file bench_report.toml.}

syntactically added this pull request to stack #1875 October 2, 2026 10:52

syntactically requested review from andreiltd, danbugs, dblnz, devigned, jprendes, jsturtevant, ludfjig, simongdavies and squillace as code owners October 2, 2026 10:52

This comment has been minimized.

Sign in to view

syntactically added kind/enhancement For PRs adding features, improving functionality, docs, tests, etc. ready-for-review PR is ready for (re-)review labels Oct 2, 2026

Copilot AI balanced review requested due to automatic review settings October 2, 2026 12:29

syntactically force-pushed the lm/extra-tlbi branch from 9717eee to e18ba90 Compare October 2, 2026 12:29

Copilot started reviewing on behalf of syntactically October 2, 2026 12:33 View session

Copilot AI reviewed Oct 2, 2026

View reviewed changes

Comment thread src/hyperlight_guest_bin/src/paging.rs

Comment thread src/hyperlight_guest_bin/src/paging.rs

Comment thread src/hyperlight_guest_bin/src/arch/aarch64/paging.rs

Comment thread src/hyperlight_guest_bin/src/arch/amd64/paging.rs

This comment has been minimized.

Sign in to view

Base automatically changed from lm/modify-mappings to main October 2, 2026 15:45

vmem: Add stronger tlb invalidation barrier

3423312

Signed-off-by: Lucy Menon <168595099+syntactically@users.noreply.github.com>

syntactically force-pushed the lm/extra-tlbi branch from e18ba90 to 3423312 Compare October 2, 2026 15:48

jprendes approved these changes Oct 2, 2026

View reviewed changes

ludfjig approved these changes Oct 2, 2026

View reviewed changes

github-actions Bot removed the ready-for-review PR is ready for (re-)review label Oct 2, 2026

Conversation

syntactically commented Oct 2, 2026

Uh oh!

This comment has been minimized.

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

This comment has been minimized.

hyperlight-gh-bot Bot commented Oct 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

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants