perf(transport): add fast_channel for high-performance packet routing (#2263)

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Ian Clarke <sanity@users.noreply.github.com>

Author: 3 people

.github/workflows/benchmarks.yml

@@ -75,32 +75,30 @@ jobs:
             echo "⚠️ No baseline found - this run will establish the baseline" >> $GITHUB_STEP_SUMMARY
           fi
 
-      # Run Level 0 benchmarks (pure logic, deterministic)
-      - name: Run Level 0 Benchmarks (Pure Logic)
-        id: bench_level0
+      # Run Transport Layer Benchmarks (actual code with mock I/O)
+      - name: Run Transport Layer Benchmarks
+        id: bench_transport
         run: |
-          echo "## Level 0: Pure Logic Benchmarks" >> $GITHUB_STEP_SUMMARY
+          echo "## Transport Layer Performance" >> $GITHUB_STEP_SUMMARY
           echo "" >> $GITHUB_STEP_SUMMARY
-          echo "These benchmarks measure pure computation without I/O:" >> $GITHUB_STEP_SUMMARY
-          echo "- AES-GCM encryption/decryption" >> $GITHUB_STEP_SUMMARY
-          echo "- Serialization" >> $GITHUB_STEP_SUMMARY
-          echo "- Nonce generation" >> $GITHUB_STEP_SUMMARY
+          echo "These benchmarks test the actual transport code with mock sockets:" >> $GITHUB_STEP_SUMMARY
+          echo "- **Message throughput**: Full pipeline (serialize → encrypt → channel → decrypt)" >> $GITHUB_STEP_SUMMARY
+          echo "- **fast_channel**: Crossbeam-based channel performance vs tokio::sync::mpsc" >> $GITHUB_STEP_SUMMARY
+          echo "- **Connection establishment**: Full handshake timing" >> $GITHUB_STEP_SUMMARY
           echo "" >> $GITHUB_STEP_SUMMARY
 
-          # Run benchmarks and capture output
-          cargo bench --bench transport_perf -- level0 2>&1 | tee bench_output.txt
+          # Run transport benchmarks (requires bench feature)
+          cargo bench --bench transport_perf --features bench -- transport 2>&1 | tee bench_output.txt
 
-          # Parse for regressions (criterion outputs "regressed" for performance decreases)
           if grep -q "regressed" bench_output.txt; then
             echo "regression_detected=true" >> $GITHUB_OUTPUT
-            echo "### ⚠️ Performance Regressions Detected" >> $GITHUB_STEP_SUMMARY
-            echo "" >> $GITHUB_STEP_SUMMARY
+            echo "### ⚠️ Transport Regressions Detected" >> $GITHUB_STEP_SUMMARY
             echo '```' >> $GITHUB_STEP_SUMMARY
-            grep -A2 "regressed" bench_output.txt >> $GITHUB_STEP_SUMMARY || true
+            grep -B5 "regressed" bench_output.txt | tail -30 >> $GITHUB_STEP_SUMMARY || true
             echo '```' >> $GITHUB_STEP_SUMMARY
           else
             echo "regression_detected=false" >> $GITHUB_OUTPUT
-            echo "### ✅ No Regressions Detected" >> $GITHUB_STEP_SUMMARY
+            echo "### ✅ No Transport Regressions" >> $GITHUB_STEP_SUMMARY
           fi
 
           # Also capture any improvements
@@ -145,25 +143,55 @@ jobs:
 
       # Post comment on PR with regression summary
       - name: Comment on PR
-        if: github.event_name == 'pull_request' && steps.bench_level0.outputs.regression_detected == 'true'
+        if: github.event_name == 'pull_request' && steps.bench_transport.outputs.regression_detected == 'true'
         uses: actions/github-script@v7
         with:
           script: |
             const fs = require('fs');
             const output = fs.readFileSync('bench_output.txt', 'utf8');
-
-            // Extract regression lines
-            const regressions = output.split('\n')
-              .filter(line => line.includes('regressed') || line.includes('change:'))
-              .slice(0, 20)  // Limit to 20 lines
-              .join('\n');
+            const lines = output.split('\n');
+
+            // Criterion outputs benchmark results like:
+            // level0/crypto/encrypt   time:   [1.234 µs 1.250 µs 1.270 µs]
+            //                         change: [+1.23% +2.34% +3.45%] (p = 0.00 < 0.01)
+            //                         Performance has regressed.
+            //
+            // We need to capture the benchmark name (line with "time:") along with regression info
+
+            let regressions = [];
+            let currentBenchmark = null;
+
+            for (let i = 0; i < lines.length; i++) {
+              const line = lines[i];
+
+              // Capture benchmark name (lines containing "time:" have the benchmark name at the start)
+              if (line.includes('time:')) {
+                currentBenchmark = line.split('time:')[0].trim();
+              }
+
+              // When we find a regression, include the benchmark name
+              if (line.includes('Performance has regressed')) {
+                // Look back for the change line
+                const changeLine = i > 0 ? lines[i - 1] : '';
+                const changeMatch = changeLine.match(/change:\s*\[([\d.+%-]+)\s+([\d.+%-]+)\s+([\d.+%-]+)\]/);
+                const changeStr = changeMatch ? changeMatch[2] : 'unknown';
+
+                if (currentBenchmark) {
+                  regressions.push(`• ${currentBenchmark}: ${changeStr}`);
+                }
+              }
+            }
+
+            const regressionsText = regressions.length > 0
+              ? regressions.slice(0, 15).join('\n')
+              : 'See workflow summary for details';
 
             const body = `## ⚠️ Performance Benchmark Regressions Detected
 
             The following benchmarks show performance regression compared to the baseline:
 
             \`\`\`
-            ${regressions || 'See workflow summary for details'}
+            ${regressionsText}
             \`\`\`
 
             > **Note:** This is informational only and does not block the PR. Please review if the regression is expected or needs investigation.

crates/core/Cargo.toml

@@ -113,6 +113,7 @@ regex = "1"
 [[bench]]
 name = "transport_perf"
 harness = false
+required-features = ["bench"]
 
 [features]
 default = ["redb", "trace", "websocket"]
@@ -123,3 +124,4 @@ websocket = ["axum/ws"]
 testing = ["freenet-stdlib/testing"]
 console-subscriber = ["dep:console-subscriber"]
 test-network = ["dep:freenet-test-network"]
+bench = []  # Exposes internal test infrastructure for benchmarking