Skip to content

crypto: coerce -0 keylen to +0 in pbkdf2 and scrypt - #63531

Closed
ljharb wants to merge 2 commits into
nodejs:mainfrom
ljharb:crypto-pbkdf2-negative-zero-keylen
Closed

ljharb wants to merge 2 commits into
nodejs:mainfrom
ljharb:crypto-pbkdf2-negative-zero-keylen

Conversation

@ljharb

@ljharb ljharb commented May 24, 2026 •

Copy link
Copy Markdown
Member

validateInt32(keylen, 'keylen', 0) lets -0 through: typeof -0 is 'number', Number.isInteger(-0) is true, and -0 < 0 is false.
The value then reaches the PBKDF2Job binding, whose IsInt32() check fails (V8 boxes -0 as a HeapNumber rather than a tagged SMI) and aborts the process with SIGABRT.
Coerce keylen to +0 after validation so the binding sees a true Int32.

Reachable from any caller that forwards a JSON-parsed value, since JSON.parse('{"keylen":-0}').keylen preserves the sign.

(an alternative implementation would be to throw on -0 - this seemed friendlier, but I'm happy to do that instead, if folks prefer)

`validateInt32(keylen, 'keylen', 0)` lets `-0` through: `typeof -0` is
`'number'`, `Number.isInteger(-0)` is `true`, and `-0 < 0` is `false`.
The value then reaches the PBKDF2Job binding, whose `IsInt32()` check
fails (V8 boxes `-0` as a HeapNumber rather than a tagged SMI) and
aborts the process with SIGABRT.
Coerce `keylen` to `+0`
after validation so the binding sees a true Int32.

Reachable from any caller that forwards a JSON-parsed value,
since `JSON.parse('{"keylen":-0}').keylen` preserves the sign.

Signed-off-by: Jordan Harband <ljharb@gmail.com>
@ljharb ljharb added the crypto Issues and PRs related to the crypto subsystem. label May 24, 2026
@nodejs-github-bot

Copy link
Copy Markdown
Collaborator

Review requested:

  • @nodejs/crypto

@nodejs-github-bot nodejs-github-bot added the needs-ci PRs that need a full CI run. label May 24, 2026
@ljharb
ljharb force-pushed the crypto-pbkdf2-negative-zero-keylen branch 2 times, most recently from e4b488e to 9700ffe Compare May 24, 2026 08:53
@panva

panva commented May 24, 2026

Copy link
Copy Markdown
Member

@ljharb can you throw in a fix for scrypt too?

@panva

panva commented May 24, 2026 •

Copy link
Copy Markdown
Member

Outside of pbkdf2 and scrypt there are also instances in cipher, diffiehellman, keygen, hash, random. I can tackle those separately in a followup if you want, or you can, up to you.

@codecov

codecov Bot commented May 24, 2026 •

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.31%. Comparing base (df09b2a) to head (c095afb).
⚠️ Report is 30 commits behind head on main.

Additional details and impacted files
@@            Coverage Diff             @@
##             main   #63531      +/-   ##
==========================================
- Coverage   90.32%   90.31%   -0.02%     
==========================================
  Files         730      730              
  Lines      234152   234156       +4     
  Branches    43900    43912      +12     
==========================================
- Hits       211499   211467      -32     
- Misses      14374    14418      +44     
+ Partials     8279     8271       -8     
Files with missing lines Coverage Δ
lib/internal/crypto/pbkdf2.js 100.00% <100.00%> (+1.52%) ⬆️
lib/internal/crypto/scrypt.js 94.89% <100.00%> (+0.07%) ⬆️

... and 31 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.

@panva

This comment was marked as outdated.

Mirror of the prior pbkdf2 fix. `validateInt32(keylen, 'keylen', 0)`
lets `-0` through (since `-0 < 0` is `false`), and the ScryptJob
binding's `IsInt32()` check at `crypto_scrypt.cc` aborts the process
with SIGABRT because V8 boxes `-0` as a HeapNumber rather than a
tagged SMI. Coerce `keylen` to `+0` after validation.
@ljharb
ljharb force-pushed the crypto-pbkdf2-negative-zero-keylen branch from 6a406f2 to c095afb Compare May 24, 2026 13:45
@panva panva added the commit-queue-rebase PRs the Commit Queue should land as multiple self-contained commits. label May 24, 2026
@panva panva changed the title crypto: coerce -0 keylen to +0 in pbkdf2 crypto: coerce -0 keylen to +0 in pbkdf2 and scrypt May 24, 2026
@panva panva added author ready PRs with CI started, the required approvals, and no outstanding review comments. request-ci Add this label to start a Jenkins CI on a PR. Only starts once the PR has an approving review. labels May 24, 2026
@github-actions github-actions Bot removed the request-ci Add this label to start a Jenkins CI on a PR. Only starts once the PR has an approving review. label May 24, 2026
@nodejs-github-bot

This comment was marked as outdated.

@nodejs-github-bot

This comment was marked as outdated.

@nodejs-github-bot

This comment was marked as outdated.

@nodejs-github-bot

Copy link
Copy Markdown
Collaborator

@panva panva added the commit-queue PRs queued for automated landing through the Commit Queue. label May 25, 2026
@akshatgit

Copy link
Copy Markdown

@ljharb — the fix covers pbkdf2/scrypt but the same crash exists in five other APIs, verified on v22.22.1:

node -e "require('crypto').hkdf('sha256','k','s','',-0,()=>{})"          # exit 134, crypto_hkdf.cc:54
node -e "require('crypto').hkdfSync('sha256','k','s','',-0)"              # exit 134, crypto_hkdf.cc:54
node -e "require('crypto').generateKeyPair('rsa',{modulusLength:-0,publicExponent:65537},()=>{})"  # exit 134, crypto_rsa.cc:137
node -e "require('crypto').generateKeyPairSync('rsa',{modulusLength:-0,publicExponent:65537})"     # exit 134, crypto_rsa.cc:137
node -e "require('crypto').createDiffieHellman(-0)"                       # exit 134, crypto_util.h:607

The += 0 coercion from this PR would work in hkdf.js, keygen.js, and dh.js. Or fixing isUint32() in validators.js once covers all callers:

function isUint32(value) {
  return value === (value >>> 0) && !Object.is(value, -0);
}

@nodejs-github-bot nodejs-github-bot added commit-queue-failed PRs whose Commit Queue landing failed and need manual intervention before retrying. and removed commit-queue PRs queued for automated landing through the Commit Queue. labels May 26, 2026
@nodejs-github-bot

Copy link
Copy Markdown
Collaborator
Commit Queue failed
- Loading data for nodejs/node/pull/63531
✔  Done loading data for nodejs/node/pull/63531
----------------------------------- PR info ------------------------------------
Title      crypto: coerce -0 keylen to +0 in pbkdf2 and scrypt (#63531)
   ⚠  Could not retrieve the email or name of the PR author's from user's GitHub profile!
Branch     ljharb:crypto-pbkdf2-negative-zero-keylen -> nodejs:main
Labels     crypto, author ready, needs-ci, commit-queue-rebase
Commits    2
 - crypto: coerce -0 keylen to +0 in pbkdf2
 - crypto: coerce -0 keylen to +0 in scrypt
Committers 1
 - Jordan Harband <ljharb@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/63531
Reviewed-By: Filip Skokan <panva.ip@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63531
Reviewed-By: Filip Skokan <panva.ip@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
--------------------------------------------------------------------------------
   ℹ  This PR was created on Sun, 24 May 2026 08:43:03 GMT
   ✔  Approvals: 3
   ✔  - Filip Skokan (@panva) (TSC): https://github.com/nodejs/node/pull/63531#pullrequestreview-4352962596
   ✔  - James M Snell (@jasnell) (TSC): https://github.com/nodejs/node/pull/63531#pullrequestreview-4353060473
   ✔  - Luigi Pinca (@lpinca): https://github.com/nodejs/node/pull/63531#pullrequestreview-4353331893
   ✔  Last GitHub CI successful
   ℹ  Last Full PR CI on 2026-05-24T20:58:13Z: https://ci.nodejs.org/job/node-test-pull-request/73682/
- Querying data for job/node-test-pull-request/73682/
✔  Build data downloaded
   ✔  Last Jenkins CI successful
--------------------------------------------------------------------------------
   ✔  No git cherry-pick in progress
   ✔  No git am in progress
   ✔  No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
 * branch                  main       -> FETCH_HEAD
✔  origin/main is now up-to-date
- Downloading patch for 63531
From https://github.com/nodejs/node
 * branch                  refs/pull/63531/merge -> FETCH_HEAD
✔  Fetched commits as c98ab680d6a0..c095afb5b868
--------------------------------------------------------------------------------
[main a4d78abcbf] crypto: coerce -0 keylen to +0 in pbkdf2
 Author: Jordan Harband <ljharb@gmail.com>
 Date: Sun May 24 10:37:02 2026 +0200
 2 files changed, 31 insertions(+)
[main 29c93964c6] crypto: coerce -0 keylen to +0 in scrypt
 Author: Jordan Harband <ljharb@gmail.com>
 Date: Sun May 24 15:40:32 2026 +0200
 2 files changed, 29 insertions(+)
   ✔  Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:352) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)

�[KExecuting: git node land --amend --yes
--------------------------------- New Message ----------------------------------
crypto: coerce -0 keylen to +0 in pbkdf2

validateInt32(keylen, 'keylen', 0) lets -0 through: typeof -0 is
'number', Number.isInteger(-0) is true, and -0 &lt; 0 is false.
The value then reaches the PBKDF2Job binding, whose IsInt32() check
fails (V8 boxes -0 as a HeapNumber rather than a tagged SMI) and
aborts the process with SIGABRT.
Coerce keylen to +0
after validation so the binding sees a true Int32.

Reachable from any caller that forwards a JSON-parsed value,
since JSON.parse('{"keylen":-0}').keylen preserves the sign.

Signed-off-by: Jordan Harband <ljharb@gmail.com>
PR-URL: #63531
Reviewed-By: Filip Skokan <panva.ip@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>

[detached HEAD 076633221f] crypto: coerce -0 keylen to +0 in pbkdf2
Author: Jordan Harband <ljharb@gmail.com>
Date: Sun May 24 10:37:02 2026 +0200
2 files changed, 31 insertions(+)
Rebasing (3/4)
Rebasing (4/4)

�[KExecuting: git node land --amend --yes
--------------------------------- New Message ----------------------------------
crypto: coerce -0 keylen to +0 in scrypt

Mirror of the prior pbkdf2 fix. validateInt32(keylen, 'keylen', 0)
lets -0 through (since -0 &lt; 0 is false), and the ScryptJob
binding's IsInt32() check at crypto_scrypt.cc aborts the process
with SIGABRT because V8 boxes -0 as a HeapNumber rather than a
tagged SMI. Coerce keylen to +0 after validation.

PR-URL: #63531
Reviewed-By: Filip Skokan <panva.ip@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>

[detached HEAD 1eabc6dadb] crypto: coerce -0 keylen to +0 in scrypt
Author: Jordan Harband <ljharb@gmail.com>
Date: Sun May 24 15:40:32 2026 +0200
2 files changed, 29 insertions(+)

�[KSuccessfully rebased and updated refs/heads/main.

✔ 076633221f525ae25eca90443b2244898aed7de7
✔ 0:0 no Assisted-by metadata assisted-by-is-trailer
✔ 0:0 no Co-authored-by metadata co-authored-by-is-trailer
✔ 0:0 skipping fixes-url fixes-url
✔ 0:0 blank line after title line-after-title
✔ 0:0 line-lengths are valid line-length
✔ 0:0 metadata is at end of message metadata-end
✔ 13:8 PR-URL is valid. pr-url
✔ 0:0 reviewers are valid reviewers
✔ 0:0 has valid Signed-off-by signed-off-by
✔ 0:0 valid subsystems subsystem
✔ 0:0 Title is formatted correctly. title-format
✔ 0:0 Title is <= 50 columns. title-length
✖ 1eabc6dadb12e85d07a015344231ce0eac3660ef
✔ 0:0 no Assisted-by metadata assisted-by-is-trailer
✔ 0:0 no Co-authored-by metadata co-authored-by-is-trailer
✔ 0:0 skipping fixes-url fixes-url
✔ 0:0 blank line after title line-after-title
✔ 0:0 line-lengths are valid line-length
✔ 0:0 metadata is at end of message metadata-end
✔ 7:8 PR-URL is valid. pr-url
✔ 0:0 reviewers are valid reviewers
✖ 0:0 Commit must have a "Signed-off-by" trailer signed-off-by
✔ 0:0 valid subsystems subsystem
✔ 0:0 Title is formatted correctly. title-format
✔ 0:0 Title is <= 50 columns. title-length

ℹ Please fix the commit message and try again.
Please manually ammend the commit message, by running
git commit --amend
Once commit message is fixed, finish the landing command running
git node land --continue

https://github.com/nodejs/node/actions/runs/26442117825

@panva

This comment was marked as resolved.

@panva panva added commit-queue-squash PRs the Commit Queue should land as one squashed commit. and removed commit-queue-failed PRs whose Commit Queue landing failed and need manual intervention before retrying. commit-queue-rebase PRs the Commit Queue should land as multiple self-contained commits. labels May 26, 2026
@panva

panva commented May 26, 2026

Copy link
Copy Markdown
Member

Landed in 3393399

panva pushed a commit that referenced this pull request May 26, 2026
`validateInt32(keylen, 'keylen', 0)` lets `-0` through: `typeof -0` is
`'number'`, `Number.isInteger(-0)` is `true`, and `-0 < 0` is `false`.
The value then reaches the PBKDF2Job binding, whose `IsInt32()` check
fails (V8 boxes `-0` as a HeapNumber rather than a tagged SMI) and
aborts the process with SIGABRT.
Coerce `keylen` to `+0`
after validation so the binding sees a true Int32.

Reachable from any caller that forwards a JSON-parsed value,
since `JSON.parse('{"keylen":-0}').keylen` preserves the sign.

Mirror of the prior pbkdf2 fix. `validateInt32(keylen, 'keylen', 0)`
lets `-0` through (since `-0 < 0` is `false`), and the ScryptJob
binding's `IsInt32()` check at `crypto_scrypt.cc` aborts the process
with SIGABRT because V8 boxes `-0` as a HeapNumber rather than a
tagged SMI. Coerce `keylen` to `+0` after validation.

Signed-off-by: Jordan Harband <ljharb@gmail.com>
PR-URL: #63531
Reviewed-By: Filip Skokan <panva.ip@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>
@panva panva closed this May 26, 2026
@ljharb

ljharb commented May 26, 2026

Copy link
Copy Markdown
Member Author

I'll do a followup for the other locations, especially since the direction is confirmed.

@panva

panva commented May 26, 2026 •

Copy link
Copy Markdown
Member

I'll do a followup for the other locations, especially since the direction is confirmed.

@ljharb #63556

@ljharb
ljharb deleted the crypto-pbkdf2-negative-zero-keylen branch May 26, 2026 15:46
ljharb added a commit to browserify/pbkdf2 that referenced this pull request May 26, 2026
`-0` slips past the `keylen < 0` guard in `precondition`
(since `-0 < 0` is `false`)
and reaches Node's native pbkdf2 binding,
whose `IsInt32()` assertion aborts the process with an uncatchable SIGABRT on Node.js 15+
(V8 holds `-0` as a double, not a tagged Int32).
Coerce it to `+0` so the binding sees a true Int32;
`-0` now behaves identically to `+0`.

Also see nodejs/node#63531