Skip to content

vfs: integrate with CJS and ESM module loaders#63653

Open
mcollina wants to merge 38 commits into
nodejs:mainfrom
mcollina:vfs-module-loader-integration
Open

vfs: integrate with CJS and ESM module loaders#63653
mcollina wants to merge 38 commits into
nodejs:mainfrom
mcollina:vfs-module-loader-integration

Conversation

@mcollina

@mcollina mcollina commented May 30, 2026

Copy link
Copy Markdown
Member

Makes require() and import resolve files served by node:vfs. Before this PR, mounted VFS files were only visible through fs.*; the loaders went straight to the real filesystem.

Design

vfs.mount() takes no arguments and returns the reserved absolute mount point of the instance: ${os.devNull}/vfs/<layerId> (for example /dev/null/vfs/0). os.devNull is a character device on POSIX and a device-namespace path on Windows; neither can have child filesystem entries, so no real path can ever exist under this root.

Everything about ownership is decidable from the path alone:

  • Dispatch: one path normalization, one prefix comparison against the namespace root, one map lookup on the layer id. O(1) in the number of mounted layers — the benchmark (benchmark/vfs/bench-fs-dispatch.js) reports flat per-call latency across 1..10 mounted layers.
  • Real-fs paths short-circuit with a single prefix check.
  • Unmount purges loader caches by prefix-scanning the mount point — no per-VFS ownership tracking is needed, and no other-VFS or real-fs entries are ever touched.
  • Symlink correctness: realpathSync of a VFS entry always resolves to another path under the same mount point, so cache entries hidden behind symlinks are captured by the prefix scan.
  • Dynamic-import identity: resolved URLs are plain file: URLs under the mount point; import(import.meta.resolve(x)) re-hits the same module job.

Two instances mounting simultaneously never collide (each gets its own layer-<id> segment), so there is no overlap validation and no ordering hazard between mounts.

Loader integration

Toggleable wrappers in the loaders. Null fast-path when no VFS is mounted; otherwise the VFS answers stat / readFile / realpath / legacyMainResolve / getFormatOfExtensionlessFile and the four package.json C++-binding calls.

Module identity follows the path: __filename, module.filename, and import.meta.url are the plain absolute path (or file: URL) of the module under the mount point — no synthetic decorations.

Review guide

Suggested reading order:

# File Role
1 lib/internal/vfs/router.js Reserved-namespace helpers: getVfsRoot, getLayerRoot, getLayerIdFromPath.
2 lib/internal/vfs/file_system.js mount() returns the layer's reserved mount point.
3 lib/internal/vfs/setup.js Heart of the PR. findVFS (O(1) lookup), fs handler, loader overrides with parity to src/node_modules.cc / src/node_file.cc, prefix-scan cache purge.
4 lib/internal/modules/helpers.js Hook surface: loader* wrappers, setLoaderFsOverrides / setLoaderPackageOverrides, purgeRealpathCacheForPrefix.
5 lib/internal/modules/cjs/loader.js stat() + TS read routed through wrappers; purgeModuleCachesForPrefix for unmount.
6 lib/internal/modules/esm/resolve.js legacyMainResolve + internalModuleStat + toRealPath routed. No URL decoration.
7 lib/internal/modules/esm/load.js getSourceSync reads via the wrapper.
8 lib/internal/modules/esm/get_format.js Extensionless format detection routed.
9 lib/internal/modules/package_json_reader.js 4 C++ binding calls routed; purgePackageJSONCacheForPrefix.
10 lib/fs.js statSync / lstatSync honour throwIfNoEntry:false on ENOENT from the VFS handler.
11 doc/api/vfs.md Mount semantics + Module loader integration section.

Tests (all gated by --experimental-vfs): test-vfs-mount, test-vfs-mount-errors, test-vfs-multi-mount, test-vfs-require, test-vfs-import, test-vfs-module-hooks, test-vfs-module-hooks-cleanup, test-vfs-package-json, test-vfs-package-json-cache, test-vfs-invalid-package-json, test-vfs-scoped-cache-purge, test-vfs-layer-id, test-vfs-layer-tag-prefix.

Refs

The reserved-namespace design follows the "no interference with valid paths in the file system" requirement from the SEA VFS requirements doc.

Out of scope

SEA + VFS, overlay/stacking of multiple VFS layers under one prefix, migrating the C++ package_configs_ cache, broader permission-model integration.

@nodejs-github-bot

Copy link
Copy Markdown
Collaborator

Review requested:

  • @nodejs/loaders

@nodejs-github-bot nodejs-github-bot added lib / src Issues and PRs related to general changes in the lib or src directory. needs-ci PRs that need a full CI run. labels May 30, 2026
@codecov

codecov Bot commented May 30, 2026

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 91.03079% with 67 lines in your changes missing coverage. Please review.
✅ Project coverage is 90.10%. Comparing base (0e2126d) to head (b993d9e).
⚠️ Report is 220 commits behind head on main.

Files with missing lines Patch % Lines
lib/internal/vfs/setup.js 86.07% 50 Missing ⚠️
src/node_modules.cc 77.96% 4 Missing and 9 partials ⚠️
lib/internal/vfs/file_system.js 96.07% 2 Missing ⚠️
lib/internal/vfs/router.js 95.45% 1 Missing and 1 partial ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main   #63653      +/-   ##
==========================================
- Coverage   92.04%   90.10%   -1.95%     
==========================================
  Files         381      741     +360     
  Lines      169208   242313   +73105     
  Branches    25926    45732   +19806     
==========================================
+ Hits       155746   218325   +62579     
- Misses      13174    15467    +2293     
- Partials      288     8521    +8233     
Files with missing lines Coverage Δ
lib/fs.js 98.36% <100.00%> (+4.13%) ⬆️
lib/internal/modules/cjs/loader.js 98.15% <100.00%> (+18.59%) ⬆️
lib/internal/modules/esm/get_format.js 95.20% <100.00%> (+20.89%) ⬆️
lib/internal/modules/esm/load.js 91.40% <100.00%> (+8.14%) ⬆️
lib/internal/modules/esm/resolve.js 99.03% <100.00%> (+12.10%) ⬆️
lib/internal/modules/helpers.js 98.75% <100.00%> (+7.96%) ⬆️
lib/internal/modules/package_json_reader.js 99.47% <100.00%> (+11.86%) ⬆️
src/node_modules.h 100.00% <ø> (ø)
lib/internal/vfs/file_system.js 99.55% <96.07%> (+0.34%) ⬆️
lib/internal/vfs/router.js 97.01% <95.45%> (+12.23%) ⬆️
... and 2 more

... and 526 files with indirect coverage changes

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
  • 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

@mcollina

Copy link
Copy Markdown
Member Author

@joyeecheung take a look, should be easier to review.

Comment thread lib/internal/modules/esm/load.js
Comment thread lib/internal/modules/esm/resolve.js
mcollina added a commit to mcollina/node that referenced this pull request Jun 1, 2026
Restore the "DO NOT depend on the patchability" warnings in esm/load.js
and esm/resolve.js that were dropped along with the fs imports. The
warning still applies; it now also points at node:vfs as one of the
formal hook mechanisms callers should reach for instead.

Addresses review feedback from @jsumners-nr in
nodejs#63653
@mcollina
mcollina force-pushed the vfs-module-loader-integration branch from 6321e08 to 51b033a Compare June 1, 2026 15:54
@mcollina mcollina added the request-ci Add this label to start a Jenkins CI on a PR. label Jun 3, 2026
@github-actions github-actions Bot removed the request-ci Add this label to start a Jenkins CI on a PR. label Jun 3, 2026
@nodejs-github-bot

Copy link
Copy Markdown
Collaborator

@nodejs-github-bot

Copy link
Copy Markdown
Collaborator

@joyeecheung joyeecheung left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

A design question recently occurred to me: have we explored the versioning of the mounting?

Comment thread lib/internal/vfs/setup.js Outdated
@mcollina

mcollina commented Jun 3, 2026

Copy link
Copy Markdown
Member Author

A design question recently occurred to me: have we explored the versioning of the mounting?

what do you mean? You mean multiple vfs layers on top of each other?

@joyeecheung

joyeecheung commented Jun 3, 2026

Copy link
Copy Markdown
Member

what do you mean? You mean multiple vfs layers on top of each other?

For the stacks to have some kind of version number/ID to identify the current status?

BTW I just noticed that there's no mention of unmount() and mount() in the VFS docs..

@mcollina

mcollina commented Jun 3, 2026

Copy link
Copy Markdown
Member Author

BTW I just noticed that there's no mention of unmount() and mount() in the VFS docs..

I did purge them when doing the splitting; I forgot to bring them back. I'll add them to this PR.

@mcollina

mcollina commented Jun 3, 2026

Copy link
Copy Markdown
Member Author

For the stacks to have some kind of version number/ID to identify the current status?

No but we totally should.

mcollina added a commit to mcollina/node that referenced this pull request Jun 4, 2026
Each VirtualFileSystem now exposes a per-process monotonically
increasing `layerId`, assigned at construction. The id is stable
across mount/unmount cycles for the lifetime of the instance and
surfaces in:

- debug() output for register / deregister so the layer stack is
  visible when NODE_DEBUG=vfs is enabled;
- the overlap ERR_INVALID_STATE message, which now names the layer
  ids of the conflicting mounts.

The id is the building block for tagging cache entries with the
owning VFS, which a follow-up will use to replace the global
loader-cache flush in deregisterVFS with a scoped purge.

Refs: nodejs#63653
Signed-off-by: Matteo Collina <hello@matteocollina.com>
@mcollina
mcollina force-pushed the vfs-module-loader-integration branch from 51b033a to 294a19c Compare June 4, 2026 07:23
mcollina added a commit to mcollina/node that referenced this pull request Jun 4, 2026
Replace the global loader-cache flush in deregisterVFS with a
scope-purge that only drops entries owned by the unmounting VFS.
Per-layer ownership is determined two ways:

- For CJS-style filename-keyed caches (Module._cache,
  Module._pathCache, the CJS stat cache, the helpers.js realpath
  cache, and the package.json caches) entries are filtered with
  `vfs.shouldHandle(filename)`. __filename stays a clean absolute
  path so user code that does `path.dirname(__filename)` or similar
  is unaffected.

- For the ESM cascaded loader's loadCache, entries are tagged at
  resolve time: when finalizeResolution() detects the resolved
  path is VFS-owned (via the new loaderGetLayerForPath hook), it
  appends `?vfs-layer=<id>` to the URL. The tag surfaces in
  `import.meta.url`, matching the cache-busting pattern used by
  HMR tooling. On deregister, entries whose URL carries the tag
  for the unmounting layer are deleted.

Multi-mount setups no longer pay the cross-VFS cache-warmup
penalty when a single VFS unmounts, and ESM modules loaded from a
VFS become reachable for purge instead of leaking forever in the
cascaded loader.

New helpers exposed for VFS:
- cjs/loader.js: clearStatCacheForVFS
- helpers.js: purgeRealpathCacheForVFS, loaderGetLayerForPath
- package_json_reader.js: purgePackageJSONCacheForVFS

Adds test-vfs-scoped-cache-purge covering both the multi-mount
isolation and the import.meta.url tag visibility.

Refs: nodejs#63653
Signed-off-by: Matteo Collina <hello@matteocollina.com>
@mcollina

mcollina commented Jun 5, 2026

Copy link
Copy Markdown
Member Author

@joyeecheung PTAL

@mcollina
mcollina force-pushed the vfs-module-loader-integration branch from 75bb5c7 to 99a5a5c Compare June 15, 2026 09:00
mcollina added 13 commits July 13, 2026 11:36
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
@mcollina
mcollina force-pushed the vfs-module-loader-integration branch from d1d72c3 to 0d5fb61 Compare July 13, 2026 13:59
@mcollina

Copy link
Copy Markdown
Member Author

@joyeecheung done

@trivikr
trivikr requested a review from joyeecheung July 15, 2026 17:27

@joyeecheung joyeecheung left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you remove the (AI generated?) comments that just restate what happens in code with prose or stating how other code calls it currently (which can get out of sync very quickly and is anti-encapsulation?)

Also I think the documentation still doesn't clarify a core question: how is the module lookup algorithm altered once an VFS is mounted. It needs to specify that precisely in the module loading spec in modules.md instead of just leaving some hand wavy descriptions like "the module loader will consult it, it will work" (but how and where in the algorithm? and when will they NOT be consulted?)

Comment thread lib/internal/modules/esm/load.js Outdated
Comment thread lib/internal/modules/helpers.js Outdated
Comment on lines +71 to +77
// Toggleable loader hooks for VFS support. Each `loader*` wrapper below
// funnels through a single factory: if an override is registered for the
// original function it is called first; anything but `undefined` short-
// circuits the wrapper, otherwise the wrapper falls through to the
// pinned native. This keeps the wrapper surface a one-liner per binding
// - adding a new binding is one `wrapLoaderMethod` call, no extra state
// variables or per-hook setter branches to keep in sync.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
// Toggleable loader hooks for VFS support. Each `loader*` wrapper below
// funnels through a single factory: if an override is registered for the
// original function it is called first; anything but `undefined` short-
// circuits the wrapper, otherwise the wrapper falls through to the
// pinned native. This keeps the wrapper surface a one-liner per binding
// - adding a new binding is one `wrapLoaderMethod` call, no extra state
// variables or per-hook setter branches to keep in sync.
// Toggleable loader hooks for VFS support.

This is just describing what obviously happens below with 7 lines of prose?

Comment thread lib/internal/modules/helpers.js Outdated
Comment on lines +673 to +674
// internal/vfs/setup.js) map directly into this array; indexes 0..6
// use packageConfig.main as the prefix, 7..9 use pkgPath directly.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
// internal/vfs/setup.js) map directly into this array; indexes 0..6
// use packageConfig.main as the prefix, 7..9 use pkgPath directly.
// internal/vfs/setup.js) map directly into this array.

This is repeating what comments below says.

Comment thread benchmark/vfs/bench-fs-dispatch.js Outdated
Comment thread doc/api/vfs.md Outdated
Comment thread doc/api/vfs.md Outdated
Comment thread doc/api/vfs.md Outdated
Comment on lines +326 to +332
participate in module resolution and loading. Both
`require()` / `require.resolve()` (CommonJS) and `import` /
`import.meta.resolve()` (ECMAScript modules) consult the VFS through
the same toggleable hooks that `node:fs` uses, so files served from
the VFS are first-class modules: `package.json` is honoured,
extensionless files are sniffed for Wasm vs. JavaScript, conditional
`exports` / `imports` work, and so on.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
participate in module resolution and loading. Both
`require()` / `require.resolve()` (CommonJS) and `import` /
`import.meta.resolve()` (ECMAScript modules) consult the VFS through
the same toggleable hooks that `node:fs` uses, so files served from
the VFS are first-class modules: `package.json` is honoured,
extensionless files are sniffed for Wasm vs. JavaScript, conditional
`exports` / `imports` work, and so on.
participate in module resolution and loading. Both
`require()` / `require.resolve()` (CommonJS) and `import` /
`import.meta.resolve()` (ECMAScript modules) consult the VFS through
the same toggleable hooks that `node:fs` uses, so files served from
the VFS are first-class modules: `package.json` is honoured,
extensionless files are sniffed for Wasm vs. JavaScript, conditional
`exports` / `imports` work, and so on.

This paragraph is listing example without actually clarifying anything - exactly how do they appear in the module lookup algorithm, and with what precedence?

mcollina added 2 commits July 18, 2026 15:10
Signed-off-by: Matteo Collina <hello@matteocollina.com>
Signed-off-by: Matteo Collina <hello@matteocollina.com>
mcollina and others added 5 commits July 24, 2026 02:34
Co-authored-by: Joyee Cheung <joyeec9h3@gmail.com>
The microbenchmark drove 1000+ iterations to reach the optimizing tier,
which is not representative of real workloads. Module-loading against a
mounted VFS is the meaningful signal and can be measured with the
existing benchmark harness.
Measures loading a large CJS or ESM module graph from a mounted VFS,
relying on the unmount cache purge so every iteration is a cold load.
@mcollina

Copy link
Copy Markdown
Member Author

@joyeecheung PTAL, I think I did all the fix you suggested.

@mcollina mcollina added the request-ci Add this label to start a Jenkins CI on a PR. label Jul 24, 2026
@github-actions github-actions Bot removed the request-ci Add this label to start a Jenkins CI on a PR. label Jul 24, 2026
@nodejs-github-bot

Copy link
Copy Markdown
Collaborator

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

lib / src Issues and PRs related to general changes in the lib or src directory. needs-ci PRs that need a full CI run. vfs Issues and PRs related to the virtual filesystem subsystem.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

7 participants