test: deflake test-benchmark-crypto

Test mode reduces bufferSize to one, giving the random buffers a 1/256
chance of matching and failing the comparison count assertion. Use 256
bytes in test mode and force the first bytes to differ before timing.
Remove the flaky designation.

Fixes: #52690
Signed-off-by: Filip Skokan <panva.ip@gmail.com>
Assisted-by: Codex
PR-URL: #66151
Fixes: #58353
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Jake Yuesong Li <jake.yuesong@gmail.com>
Reviewed-By: Gürgün Dayıoğlu <hey@gurgun.day>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Richard Lau <richard.lau@ibm.com>
Reviewed-By: Luigi Pinca <luigipinca@gmail.com>

Author: panva

benchmark/crypto/timingSafeEqual.js

@@ -7,10 +7,14 @@ const { randomBytes, timingSafeEqual } = require('node:crypto');
 const bench = common.createBenchmark(main, {
   n: [1e5],
   bufferSize: [10, 100, 200, 2_100, 22_023],
+}, {
+  test: { bufferSize: 256 },
 });
 
 function main({ n, bufferSize }) {
   const bufs = [randomBytes(bufferSize), randomBytes(bufferSize)];
+  // Ensure the buffers differ even if the random bytes are identical.
+  bufs[1][0] = bufs[0][0] ^ 1;
   bench.start();
   let count = 0;
   for (let i = 0; i < n; i++) {

test/benchmark/benchmark.status

@@ -6,9 +6,6 @@ prefix benchmark
 
 [true] # This section applies to all platforms
 
-# https://github.com/nodejs/node/issues/52690
-test-benchmark-crypto: PASS, FLAKY
-
 [$system==win32]
 
 [$system==linux]