crypto: reduce EC key processing overhead by panva · Pull Request #65908 · nodejs/node

panva · 2026-09-08T10:17:16Z

Reduces unnecessary work when importing and using EC keys.

Streamlines private JWK import while preserving validation, and reads curve metadata, JWK coordinates, raw key material, and signature component sizes directly instead of reconstructing EC keys. Includes focused tests and benchmarks.

~~Raw private-key imports defer public-point derivation until needed, allowing signing and ECDH to avoid that cost. Operations needing the public point derive it once and cache the completed key.~~ Dropped because this optimization does not work on OpenSSL >= 4 openssl/openssl#29054 makes provider imports derive the missing public point eagerly. rip 0ddf3c9

Avoid creating temporary EVP objects and repeating key validation while retaining the private scalar range and public/private consistency checks. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

Avoid reconstructing EC keys for key details and TLS ephemeral-key curve reporting. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

Query provider coordinates together instead of serializing and decoding the public point. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

Read private scalars and matching uncompressed provider encodings directly. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

nodejs-github-bot · 2026-09-08T10:17:22Z

Review requested:

@nodejs/crypto
@nodejs/performance
@nodejs/security-wg

Use EVP_PKEY_bits() to determine the width of ECDSA signature components on OpenSSL 3. This avoids reconstructing the EC group and public point just to read the group order size. Signed-off-by: Filip Skokan <panva.ip@gmail.com> Assisted-by: GitHub Copilot

github-actions · 2026-09-08T11:11:58Z

Benchmark GHA (crypto / create-keyobject): https://github.com/nodejs/node/actions/runs/34219400423

Results

[!WARNING]
Do not take GHA benchmark results as face value, always confirm them
using a dedicated machine, e.g. Jenkins CI.

Benchmark results:

                                                                              confidence improvement accuracy (*)    (**)   (***)
crypto/create-keyobject.js n=1000 keyFormat='der-pkcs8' keyType='ec'                         -0.82 %       ±8.08% ±10.65% ±13.67%
crypto/create-keyobject.js n=1000 keyFormat='der-pkcs8' keyType='ed25519'                    -0.80 %       ±6.67%  ±8.80% ±11.29%
crypto/create-keyobject.js n=1000 keyFormat='der-pkcs8' keyType='ml-dsa-44'                   1.95 %       ±5.51%  ±7.27%  ±9.33%
crypto/create-keyobject.js n=1000 keyFormat='der-pkcs8' keyType='rsa'                        -1.72 %       ±7.75% ±10.22% ±13.12%
crypto/create-keyobject.js n=1000 keyFormat='der-spki' keyType='ec'                           1.97 %       ±8.61% ±11.35% ±14.56%
crypto/create-keyobject.js n=1000 keyFormat='der-spki' keyType='ed25519'                      0.52 %       ±8.44% ±11.13% ±14.28%
crypto/create-keyobject.js n=1000 keyFormat='der-spki' keyType='ml-dsa-44'                    0.45 %       ±9.07% ±11.96% ±15.35%
crypto/create-keyobject.js n=1000 keyFormat='der-spki' keyType='rsa'                         -0.57 %       ±8.60% ±11.33% ±14.54%
crypto/create-keyobject.js n=1000 keyFormat='jwk-private' keyType='ec'               ***     46.97 %       ±8.21% ±10.82% ±13.89%
crypto/create-keyobject.js n=1000 keyFormat='jwk-private' keyType='ed25519'                  -1.56 %       ±6.09%  ±8.02% ±10.30%
crypto/create-keyobject.js n=1000 keyFormat='jwk-private' keyType='ml-dsa-44'                -0.94 %       ±5.22%  ±6.87%  ±8.82%
crypto/create-keyobject.js n=1000 keyFormat='jwk-private' keyType='rsa'                      -0.50 %       ±6.39%  ±8.42% ±10.80%
crypto/create-keyobject.js n=1000 keyFormat='jwk-public' keyType='ec'                        -0.38 %       ±7.52%  ±9.91% ±12.72%
crypto/create-keyobject.js n=1000 keyFormat='jwk-public' keyType='ed25519'                   -0.09 %       ±8.04% ±10.60% ±13.60%
crypto/create-keyobject.js n=1000 keyFormat='jwk-public' keyType='ml-dsa-44'                  0.46 %       ±7.87% ±10.37% ±13.30%
crypto/create-keyobject.js n=1000 keyFormat='jwk-public' keyType='rsa'                       -0.19 %       ±7.71% ±10.16% ±13.04%
crypto/create-keyobject.js n=1000 keyFormat='pkcs8' keyType='ec'                             -1.33 %       ±8.14% ±10.73% ±13.76%
crypto/create-keyobject.js n=1000 keyFormat='pkcs8' keyType='ed25519'                         0.45 %       ±6.58%  ±8.68% ±11.13%
crypto/create-keyobject.js n=1000 keyFormat='pkcs8' keyType='ml-dsa-44'                       0.52 %       ±5.57%  ±7.34%  ±9.42%
crypto/create-keyobject.js n=1000 keyFormat='pkcs8' keyType='rsa'                            -1.64 %       ±7.07%  ±9.33% ±11.97%
crypto/create-keyobject.js n=1000 keyFormat='raw-private' keyType='ec'                       -1.86 %       ±6.41%  ±8.46% ±10.85%
crypto/create-keyobject.js n=1000 keyFormat='raw-private' keyType='ed25519'                  -0.65 %       ±6.43%  ±8.48% ±10.88%
crypto/create-keyobject.js n=1000 keyFormat='raw-public' keyType='ec'                        -1.35 %       ±7.30%  ±9.63% ±12.35%
crypto/create-keyobject.js n=1000 keyFormat='raw-public' keyType='ed25519'                   -1.38 %       ±8.14% ±10.73% ±13.76%
crypto/create-keyobject.js n=1000 keyFormat='raw-public' keyType='ml-dsa-44'                  0.91 %       ±9.13% ±12.03% ±15.44%
crypto/create-keyobject.js n=1000 keyFormat='raw-seed' keyType='ml-dsa-44'                   -0.30 %       ±5.49%  ±7.24%  ±9.28%
crypto/create-keyobject.js n=1000 keyFormat='spki' keyType='ec'                               0.01 %       ±8.48% ±11.18% ±14.34%
crypto/create-keyobject.js n=1000 keyFormat='spki' keyType='ed25519'                          0.39 %       ±8.56% ±11.28% ±14.48%
crypto/create-keyobject.js n=1000 keyFormat='spki' keyType='ml-dsa-44'                        0.21 %       ±7.72% ±10.18% ±13.06%
crypto/create-keyobject.js n=1000 keyFormat='spki' keyType='rsa'                              1.12 %       ±8.38% ±11.05% ±14.18%

Be aware that when doing many comparisons the risk of a false-positive
result increases. In this case, there are 30 comparisons, you can thus
expect the following amount of false-positive results:
  1.50 false positives, when considering a   5% risk acceptance (*, **, ***),
  0.30 false positives, when considering a   1% risk acceptance (**, ***),
  0.03 false positives, when considering a 0.1% risk acceptance (***)

[!WARNING]
Do not take GHA benchmark results as face value, always confirm them
using a dedicated machine, e.g. Jenkins CI.

github-actions · 2026-09-08T11:12:02Z

Benchmark GHA (crypto / ec-): https://github.com/nodejs/node/actions/runs/34219408729

Results

[!WARNING]
Do not take GHA benchmark results as face value, always confirm them
using a dedicated machine, e.g. Jenkins CI.

Benchmark results:

                                                                                                 confidence improvement accuracy (*)    (**)    (***)
crypto/ec-jwk-export.js n=10000 type='private' namedCurve='prime256v1'                                  ***    134.04 %      ±11.18% ±14.76%  ±18.98%
crypto/ec-jwk-export.js n=10000 type='private' namedCurve='secp256k1'                                   ***    238.46 %      ±19.90% ±26.29%  ±33.85%
crypto/ec-jwk-export.js n=10000 type='private' namedCurve='secp384r1'                                   ***    224.90 %      ±45.03% ±59.48%  ±76.57%
crypto/ec-jwk-export.js n=10000 type='private' namedCurve='secp521r1'                                   ***    243.52 %      ±44.67% ±59.02%  ±75.99%
crypto/ec-jwk-export.js n=10000 type='public' namedCurve='prime256v1'                                   ***    161.24 %      ±12.20% ±16.10%  ±20.71%
crypto/ec-jwk-export.js n=10000 type='public' namedCurve='secp256k1'                                    ***    301.64 %      ±24.63% ±32.54%  ±41.90%
crypto/ec-jwk-export.js n=10000 type='public' namedCurve='secp384r1'                                    ***    282.24 %      ±54.68% ±72.24%  ±93.03%
crypto/ec-jwk-export.js n=10000 type='public' namedCurve='secp521r1'                                    ***    326.33 %      ±57.90% ±76.51%  ±98.55%
crypto/ec-key-details.js n=10000 type='private' namedCurve='P-256'                                      ***    318.72 %      ±18.62% ±24.60%  ±31.68%
crypto/ec-key-details.js n=10000 type='private' namedCurve='P-384'                                      ***    516.83 %      ±29.72% ±39.23%  ±50.47%
crypto/ec-key-details.js n=10000 type='private' namedCurve='P-521'                                      ***    568.32 %      ±32.94% ±43.52%  ±56.02%
crypto/ec-key-details.js n=10000 type='public' namedCurve='P-256'                                       ***    318.14 %      ±19.15% ±25.30%  ±32.59%
crypto/ec-key-details.js n=10000 type='public' namedCurve='P-384'                                       ***    507.42 %      ±29.64% ±39.13%  ±50.34%
crypto/ec-key-details.js n=10000 type='public' namedCurve='P-521'                                       ***    562.96 %      ±31.38% ±41.45%  ±53.35%
crypto/ec-raw-export.js n=10000 type='compressed' format='raw-public' namedCurve='prime256v1'                    0.37 %       ±4.75%  ±6.26%   ±8.03%
crypto/ec-raw-export.js n=10000 type='compressed' format='raw-public' namedCurve='secp384r1'                     0.96 %      ±17.89% ±23.58%  ±30.26%
crypto/ec-raw-export.js n=10000 type='compressed' format='raw-public' namedCurve='secp521r1'                    -0.40 %      ±15.80% ±20.83%  ±26.73%
crypto/ec-raw-export.js n=10000 type='uncompressed' format='raw-private' namedCurve='prime256v1'        ***    117.92 %       ±7.90% ±10.43%  ±13.40%
crypto/ec-raw-export.js n=10000 type='uncompressed' format='raw-private' namedCurve='secp384r1'         ***    216.74 %      ±14.69% ±19.37%  ±24.86%
crypto/ec-raw-export.js n=10000 type='uncompressed' format='raw-private' namedCurve='secp521r1'         ***    252.17 %      ±15.10% ±19.91%  ±25.55%
crypto/ec-raw-export.js n=10000 type='uncompressed' format='raw-public' namedCurve='prime256v1'         ***     93.30 %       ±5.70%  ±7.51%   ±9.65%
crypto/ec-raw-export.js n=10000 type='uncompressed' format='raw-public' namedCurve='secp384r1'          ***    287.38 %      ±62.04% ±81.99% ±105.60%
crypto/ec-raw-export.js n=10000 type='uncompressed' format='raw-public' namedCurve='secp521r1'          ***    302.08 %      ±63.06% ±83.33% ±107.35%

Be aware that when doing many comparisons the risk of a false-positive
result increases. In this case, there are 23 comparisons, you can thus
expect the following amount of false-positive results:
  1.15 false positives, when considering a   5% risk acceptance (*, **, ***),
  0.23 false positives, when considering a   1% risk acceptance (**, ***),
  0.02 false positives, when considering a 0.1% risk acceptance (***)

[!WARNING]
Do not take GHA benchmark results as face value, always confirm them
using a dedicated machine, e.g. Jenkins CI.

github-actions · 2026-09-08T11:12:07Z

Benchmark GHA (crypto / webcrypto-sign): https://github.com/nodejs/node/actions/runs/34219404291

Results

[!WARNING]
Do not take GHA benchmark results as face value, always confirm them
using a dedicated machine, e.g. Jenkins CI.

Benchmark results:

                                                                                              confidence improvement accuracy (*)    (**)   (***)
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='parallel' keyType='ec'                       ***     18.15 %       ±8.92% ±11.76% ±15.10%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='parallel' keyType='ed25519'                           0.21 %       ±7.50%  ±9.89% ±12.69%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='parallel' keyType='ml-dsa-44'                         1.17 %       ±8.24% ±10.86% ±13.94%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='parallel' keyType='rsa-pss'                          -1.83 %       ±6.62%  ±8.73% ±11.20%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='parallel' keyType='rsassa-pkcs1-v1_5'                -1.32 %       ±6.91%  ±9.11% ±11.68%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='serial' keyType='ec'                         ***     17.72 %       ±9.00% ±11.86% ±15.22%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='serial' keyType='ed25519'                            -2.12 %       ±7.46%  ±9.83% ±12.62%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='serial' keyType='ml-dsa-44'                           0.47 %       ±6.88%  ±9.06% ±11.63%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='serial' keyType='rsa-pss'                            -2.23 %       ±3.79%  ±5.00%  ±6.42%
crypto/webcrypto-sign.js n=1000 keyReuse='shared' mode='serial' keyType='rsassa-pkcs1-v1_5'                  -0.33 %       ±3.62%  ±4.77%  ±6.13%
crypto/webcrypto-sign.js n=1000 keyReuse='unique' mode='parallel' keyType='ec'                       ***     15.75 %       ±8.80% ±11.60% ±14.90%
crypto/webcrypto-sign.js n=1000 keyReuse='unique' mode='parallel' keyType='ed25519'                           1.41 %       ±7.24%  ±9.54% ±12.24%
crypto/webcrypto-sign.js n=1000 keyReuse='unique' mode='parallel' keyType='ml-dsa-44'                         0.15 %       ±8.00% ±10.54% ±13.52%
crypto/webcrypto-sign.js n=1000 keyReuse='unique' mode='parallel' keyType='rsa-pss'                          -2.76 %       ±6.05%  ±7.98% ±10.24%
crypto/webcrypto-sign.js n=1000 keyReuse='unique' mode='parallel' keyType='rsassa-pkcs1-v1_5'                -2.97 %       ±6.05%  ±7.98% ±10.24%

Be aware that when doing many comparisons the risk of a false-positive
result increases. In this case, there are 15 comparisons, you can thus
expect the following amount of false-positive results:
  0.75 false positives, when considering a   5% risk acceptance (*, **, ***),
  0.15 false positives, when considering a   1% risk acceptance (**, ***),
  0.01 false positives, when considering a 0.1% risk acceptance (***)

[!WARNING]
Do not take GHA benchmark results as face value, always confirm them
using a dedicated machine, e.g. Jenkins CI.

codecov · 2026-09-08T13:42:55Z

Codecov Report

❌ Patch coverage is 81.08108% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 90.19%. Comparing base (70458d5) to head (f349ae1).
⚠️ Report is 5 commits behind head on main.

Files with missing lines	Patch %	Lines
src/crypto/crypto_ec.cc	71.42%	2 Missing and 2 partials ⚠️
src/crypto/crypto_keys.cc	86.36%	1 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #65908      +/-   ##
==========================================
+ Coverage   90.18%   90.19%   +0.01%     
==========================================
  Files         771      771              
  Lines      265103   265039      -64     
  Branches    50353    50333      -20     
==========================================
- Hits       239082   239061      -21     
+ Misses      16968    16943      -25     
+ Partials     9053     9035      -18

Files with missing lines	Coverage Δ
src/crypto/crypto_common.cc	`78.12% <100.00%> (+0.80%)`	⬆️
src/crypto/crypto_keys.cc	`70.47% <86.36%> (+0.81%)`	⬆️
src/crypto/crypto_ec.cc	`69.75% <71.42%> (+1.33%)`	⬆️

... and 27 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 added 4 commits September 8, 2026 12:00

crypto: optimize private EC JWK import

e9c84d6

Avoid creating temporary EVP objects and repeating key validation while retaining the private scalar range and public/private consistency checks. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

crypto: read EC curve metadata directly

faa4288

Avoid reconstructing EC keys for key details and TLS ephemeral-key curve reporting. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

crypto: export EC JWK coordinates directly

ad91124

Query provider coordinates together instead of serializing and decoding the public point. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

crypto: avoid EC raw export reconstruction

842794d

Read private scalars and matching uncompressed provider encodings directly. Assisted-by: GitHub Copilot Signed-off-by: Filip Skokan <panva.ip@gmail.com>

panva added performance Issues and PRs related to the performance of Node.js. commit-queue-rebase PRs the Commit Queue should land as multiple self-contained commits. labels Sep 8, 2026

nodejs-github-bot added lib / src Issues and PRs involving general changes in the lib/ or src/ directories. needs-ci PRs that need a full CI run. labels Sep 8, 2026

panva marked this pull request as draft September 8, 2026 10:31

nodejs deleted a comment from github-actions Bot Sep 8, 2026

panva force-pushed the crypto-ec-raw-private-deferred branch from b534fc9 to b406b68 Compare September 8, 2026 10:56

panva changed the title ~~crypto: reduce EC key import and export overhead~~ crypto: reduce EC key processing overhead Sep 8, 2026

panva marked this pull request as ready for review September 8, 2026 11:01

panva added the crypto Issues and PRs related to the crypto subsystem. label Sep 8, 2026

fixup! crypto: export EC JWK coordinates directly

f349ae1

This was referenced Sep 8, 2026

crypto: optimize getBytesOfRS for EC keys #65889

Closed

crypto: optimize private EC JWK import #65893

Closed

panva requested a review from jasnell September 8, 2026 13:36

Navigation Menu

Uh oh!

Uh oh!

crypto: reduce EC key processing overhead - #65908

crypto: reduce EC key processing overhead#65908
panva wants to merge 6 commits into
nodejs:mainfrom
panva:crypto-ec-raw-private-deferred

panva commented Sep 8, 2026 •

edited

Loading

Uh oh!

nodejs-github-bot commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Sep 8, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Sep 8, 2026 •

edited

Loading

Uh oh!

codecov Bot commented Sep 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Uh oh!

Conversation

panva commented Sep 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nodejs-github-bot commented Sep 8, 2026

Uh oh!

github-actions Bot commented Sep 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Sep 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Sep 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov Bot commented Sep 8, 2026

Codecov Report

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

panva commented Sep 8, 2026 •

edited

Loading

github-actions Bot commented Sep 8, 2026 •

edited

Loading

github-actions Bot commented Sep 8, 2026 •

edited

Loading

github-actions Bot commented Sep 8, 2026 •

edited

Loading