Skip to content

child_process: reject invalid environment variable names - #67659

Open
ShogunPanda wants to merge 1 commit into
nodejs:mainfrom
ShogunPanda:child-process-reject-invalid-env-keys
Open

ShogunPanda wants to merge 1 commit into
nodejs:mainfrom
ShogunPanda:child-process-reject-invalid-env-keys

Conversation

@ShogunPanda

Copy link
Copy Markdown
Contributor

child_process.spawn() and friends build the child environment as
KEY=VALUE strings. A key in options.env that is empty or that
contains = cannot be represented that way: for example
{ 'A=B': 'C' } produced A=B=C, which the child reads as variable A
with value B=C.

This throws ERR_INVALID_ARG_VALUE for such keys in
normalizeSpawnArguments(), next to the existing null byte check. On
Windows a leading = is still accepted, because process.env contains
per-drive entries like =C: and env: process.env must keep working.
On POSIX, process.env does not expose entries starting with =, so
env: process.env is unaffected.

This is a bug fix and not an API change: such input has never been
representable, and the child always received a different variable than
the one requested. The same applies to the null byte validation of
env, which landed without a semver label and was backported to
v18.x and v19.x (#44782).

Includes a test covering spawn, spawnSync, exec, execSync,
execFile, execFileSync and fork, a positive case for a leading =
on Windows, and a one line doc note.

Keys in the `env` option that are empty or that contain `=` cannot be
represented in the `KEY=VALUE` form used by the child process
environment. They were passed through unchanged, so the child received
a different variable than the one the caller wrote. Throw
`ERR_INVALID_ARG_VALUE` for them, as is already done for keys that
contain null bytes.

On Windows, a leading `=` is allowed, as it is used by per-drive
variables such as `=C:`.

Signed-off-by: Paolo Insogna <paolo@cowtech.it>
@nodejs-github-bot nodejs-github-bot added child_process Issues and PRs related to the child_process subsystem. needs-ci PRs that need a full CI run. labels Nov 20, 2026
invalidKeys.push('=C:');
} else {
// A lone '=' is not a valid name on Windows either.
invalidKeys.push('=');

Copy link
Copy Markdown

Choose a reason for hiding this comment

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

I think it might be worth testing the lone = unconditionally, not just for Windows.

@codecov

codecov Bot commented Nov 20, 2026

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 85.71429% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 90.73%. Comparing base (030745c) to head (5285e5b).
⚠️ Report is 2 commits behind head on main.

Files with missing lines Patch % Lines
lib/child_process.js 85.71% 0 Missing and 1 partial ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main   #67659      +/-   ##
==========================================
+ Coverage   90.72%   90.73%   +0.01%     
==========================================
  Files         801      801              
  Lines      280960   280967       +7     
  Branches    54260    54270      +10     
==========================================
+ Hits       254892   254932      +40     
+ Misses      16248    16199      -49     
- Partials     9820     9836      +16     
Files with missing lines Coverage Δ
lib/child_process.js 91.51% <85.71%> (-0.04%) ⬇️

... and 34 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.

@sepehrst

sepehrst commented Nov 20, 2026 •

Copy link
Copy Markdown

On Windows, in addition to the lone = also anything starting with == is invalid
Non-standard drive letters can be created and used in CMD (for example, subst #: C:\ creates #:), whereas while creating =: succeeds it fails to retain the current directory information as Windows cannot create the required environment variable ==:.
So it might be worth also excluding anything starting with == on Windows:

[C:\]
$ subst ^=: C:\

[C:\]
$ ^=:

[=:\]
$ cd Windows
Not enough memory resources are available to process this command.

[=:\Windows]
$ C:

[C:\]
$ ^=:

[=:\]
$

For comparison:

[C:\]
$ subst #: C:\

[C:\]
$ #:

[#:\]
$ cd Windows

[#:\Windows]
$ C:

[C:\]
$ #:

[#:\Windows]
$

Edit:
Just noticed StringPrototypeIndexOf(key, '=', isWindows ? 1 : 0) !== -1 already rejects anything starting with ==. So the whole comment might be irrelevant, but I kept it anyway just in case adding an explicit test case for ^== (especially on Windows) seems worthwhile to guard against future refactors.

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

Labels

child_process Issues and PRs related to the child_process subsystem. needs-ci PRs that need a full CI run.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants