vmem: Add function for modifying mappings in-place by syntactically · Pull Request #1874 · hyperlight-dev/hyperlight

syntactically · 2026-10-02T10:51:18Z

This is useful for things like changing the permissions on an existing mapping.

This is useful for things like changing the permissions on an existing mapping. Signed-off-by: Lucy Menon <168595099+syntactically@users.noreply.github.com>

Copilot

Copilot review overview

🟡 Changes recommended

Both implementations allocate page tables for unmapped holes, risking guest scratch exhaustion.

Review effort: Balanced
Findings: 2 · 3

Open (5)

Skip unmapped subtrees instead of allocating intermediate tables · New
Avoid allocating page tables for absent subtrees · New
Add tests for mapping mutations and sparse ranges · New
Document the unsafe function's safety requirements · New
Correct the TLB invalidation safety condition · New

What changed in this PR

Adds in-place page-table mapping updates for guest memory permissions.

Changes:

Exposes modify_mapping through common and guest paging APIs.
Implements mapping updates for amd64 and AArch64.
Reuses AArch64 descriptor decoding.

File	Description
`src/hyperlight_guest_bin/src/paging.rs`	Re-exports the guest API.
`src/hyperlight_guest_bin/src/arch/amd64/paging.rs`	Adds the amd64 wrapper.
`src/hyperlight_guest_bin/src/arch/aarch64/paging.rs`	Adds the AArch64 wrapper.
`src/hyperlight_common/src/vmem.rs`	Documents and exports the common API.
`src/hyperlight_common/src/arch/amd64/vmem.rs`	Implements amd64 updates.
`src/hyperlight_common/src/arch/aarch64/vmem.rs`	Implements AArch64 updates and shares descriptor decoding.

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

hyperlight-gh-bot · 2026-10-02T13:00:50Z

Benchmark Results

Measured commit: 33a6193dc8a7
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`	`787.20 ns` (➖ 1.02x slower)
`vec_bytes`	`581.89 ns` (➖ 1.00x faster)
		`373.94 µs` (➖ 1.00x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`530.98 ns` (➖ 1.01x faster)
`65536`	`175.80 ns` (➖ 1.45x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`84.35 ms` (➖ 1.06x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`7.83 ns` (➖ 1.03x slower)	`8.21 ns` (➖ 1.04x slower)	`8.04 ns` (➖ 1.03x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`93.00 µs` (➖ 1.03x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.46 µs` (➖ 1.03x slower)	`7.40 µs` (➖ 1.01x faster)
`65536`	`2.10 µs` (➖ 1.01x slower)	`2.01 µs` (➖ 1.03x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`6.58 µs` (➖ 1.05x slower)	`6.45 µs` (➖ 1.03x slower)
`8192`	`1.11 µs` (➖ 1.10x slower)	`1.07 µs` (➖ 1.03x slower)
`262144`		`28.59 µs` (➖ 1.07x slower)

kvm / intel (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`786.67 ns` (➖ 1.05x slower)
`vec_bytes`	`600.21 ns` (➖ 1.10x slower)
		`747.82 µs` (➖ 1.08x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`608.87 ns` (➖ 1.06x slower)
`65536`	`164.28 ns` (➖ 1.07x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`81.43 ms` (➖ 1.03x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`8.28 ns` (➖ 1.10x slower)	`8.28 ns` (➖ 1.11x slower)	`8.28 ns` (➖ 1.10x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`51.88 µs` (➖ 1.06x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`8.93 µs` (➖ 1.04x slower)	`8.90 µs` (➖ 1.04x slower)
`65536`	`2.50 µs` (➖ 1.04x slower)	`2.61 µs` (➖ 1.09x slower)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.35 µs` (➖ 1.06x slower)	`8.35 µs` (➖ 1.05x slower)
`8192`	`874.64 ns` (➖ 1.04x slower)	`836.07 ns` (➖ 1.06x slower)
`262144`		`35.56 µs` (➖ 1.06x slower)

mshv3 / amd (Linux) (➖ stable)

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`951.04 ns` (➖ 1.00x faster)
`vec_bytes`	`708.87 ns` (➖ 1.00x faster)
		`271.00 µs` (➖ 1.31x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`721.78 ns` (➖ 1.02x faster)
`65536`	`192.08 ns` (➖ 1.01x faster)

sandboxes

	`create_initialized_and_drop`
`medium`	`58.06 ms` (➖ 1.04x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.75 ns` (➖ 1.00x faster)	`9.74 ns` (➖ 1.00x slower)	`10.12 ns` (➖ 1.01x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`85.27 µs` (➖ 1.02x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`10.13 µs` (➖ 1.00x slower)	`8.79 µs` (➖ 1.02x faster)
`65536`	`2.24 µs` (➖ 1.01x slower)	`2.48 µs` (➖ 1.01x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.41 µs` (➖ 1.03x slower)	`8.48 µs` (➖ 1.05x slower)
`8192`	`1.27 µs` (➖ 1.00x slower)	`1.32 µs` (➖ 1.01x faster)
`262144`		`36.51 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`	`937.22 ns` (➖ 1.00x slower)
`vec_bytes`	`688.52 ns` (➖ 1.02x slower)
		`991.64 µs` (➖ 1.44x slower)

payload_allocation

	`slot_pool_segmented`
`262144`	`635.10 ns` (➖ 1.01x faster)
`65536`	`166.30 ns` (➖ 1.01x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`68.48 ms` (➖ 1.10x slower)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.31 ns` (➖ 1.03x slower)	`8.79 ns` (➖ 1.01x slower)	`8.86 ns` (➖ 1.01x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`47.51 µs` (➖ 1.09x slower)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.80 µs` (➖ 1.01x slower)	`7.78 µs` (➖ 1.03x faster)
`65536`	`2.23 µs` (➖ 1.00x faster)	`2.27 µs` (➖ 1.00x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`7.57 µs` (➖ 1.01x slower)	`7.62 µs` (➖ 1.01x slower)
`8192`	`877.69 ns` (➖ 1.03x slower)	`892.35 ns` (➖ 1.01x slower)
`262144`		`37.77 µs` (➖ 1.00x faster)

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

No benchmark improved or regressed.

Benchmark Results

function_call_codec

	`encode_control`	`decode_vec_bytes_copy`
`byte_chunks`	`1.16 µs` (➖ 1.05x slower)
`vec_bytes`	`808.86 ns` (➖ 1.20x faster)
		`2.16 ms` (➖ 1.16x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`787.76 ns` (➖ 1.01x faster)
`65536`	`230.98 ns` (➖ 1.01x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`86.79 ms` (➖ 1.27x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`9.97 ns` (➖ 1.39x faster)	`10.25 ns` (➖ 1.03x faster)	`10.14 ns` (➖ 1.04x faster)

snapshot_files

	`load_snapshot_unverified`
`small`	`701.38 µs` (➖ 1.13x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`9.14 µs` (➖ 1.05x slower)	`9.24 µs` (➖ 1.01x faster)
`65536`	`2.40 µs` (➖ 1.06x faster)	`2.38 µs` (➖ 1.13x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`9.02 µs` (➖ 1.05x slower)	`8.64 µs` (➖ 1.01x faster)
`8192`	`1.31 µs` (➖ 1.09x faster)	`1.33 µs` (➖ 1.06x faster)
`262144`		`45.83 µs` (➖ 1.18x 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 faster)
`vec_bytes`	`801.37 ns` (➖ 1.04x slower)
		`3.05 ms` (➖ 1.02x faster)

payload_allocation

	`slot_pool_segmented`
`262144`	`755.07 ns` (➖ 1.02x slower)
`65536`	`211.61 ns` (➖ 1.01x slower)

sandboxes

	`create_initialized_and_drop`
`medium`	`109.58 ms` (➖ 1.10x faster)

slot_pool

	`alloc_dealloc_1500`	`alloc_dealloc_4096`	`alloc_dealloc_128`
	`10.20 ns` (➖ 1.03x slower)	`10.57 ns` (➖ 1.05x slower)	`10.89 ns` (➖ 1.02x slower)

snapshot_files

	`load_snapshot_unverified`
`small`	`602.67 µs` (➖ 1.02x faster)

virtq_readonly

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`262144`	`7.78 µs` (➖ 1.02x slower)	`7.70 µs` (➖ 1.01x faster)
`65536`	`2.33 µs` (➖ 1.03x slower)	`2.31 µs` (➖ 1.00x faster)

virtq_readwrite

	`slot_pool_segmented_fragmented`	`slot_pool_segmented`
`65536`	`8.62 µs` (➖ 1.04x slower)	`8.47 µs` (➖ 1.00x faster)
`8192`	`1.15 µs` (➖ 1.09x faster)	`1.22 µs` (➖ 1.09x slower)
`262144`		`60.27 µs` (➖ 1.40x slower)

_{Reported by cargo ci bench-report --candidate run:37007006191 --baseline run:36945564806 --config-file bench_report.toml.}

syntactically added this pull request to stack #1875 October 2, 2026 10:51

syntactically requested a review from danbugs as a code owner October 2, 2026 10:51

syntactically added the kind/enhancement For PRs adding features, improving functionality, docs, tests, etc. label Oct 2, 2026

syntactically requested review from dblnz and devigned as code owners October 2, 2026 10:51

syntactically added the ready-for-review PR is ready for (re-)review label Oct 2, 2026

syntactically requested review from andreiltd, jprendes, jsturtevant, ludfjig, simongdavies and squillace as code owners October 2, 2026 10:51

This comment has been minimized.

Sign in to view

vmem: Add function for modifying mappings in-place

33a6193

This is useful for things like changing the permissions on an existing mapping. Signed-off-by: Lucy Menon <168595099+syntactically@users.noreply.github.com>

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

syntactically force-pushed the lm/modify-mappings branch from e277b65 to 33a6193 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

Base automatically changed from lm/zero-init-rgns to main October 2, 2026 14:55

andreiltd approved these changes Oct 2, 2026

View reviewed changes

syntactically merged commit 33a6193 into main Oct 2, 2026
72 of 78 checks passed

syntactically deleted the lm/modify-mappings branch October 2, 2026 15:45

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!

Uh oh!

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!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants