lib: bound the generated source map cache

doc/api/module.md

@@ -1924,6 +1924,10 @@ added:
 `path` is the resolved path for the file for which a corresponding source map
 should be fetched.
 
+Source maps of code generated by `eval` or `new Function` are held in a cache of
+limited size, so the least recently used of them are dropped once enough
+generated code has been evaluated.
+
 ### `module.setSourceMapsSupport(enabled[, options])`
 
 <!-- YAML

lib/internal/source_map/source_map_cache.js

@@ -1,6 +1,7 @@
 'use strict';
 
 const {
+  ArrayIsArray,
   ArrayPrototypeIndexOf,
   ArrayPrototypePush,
   JSONParse,
@@ -36,11 +37,17 @@ const getModuleSourceMapCache = getLazy(() => {
   return new SourceMapCacheMap();
 });
 
-// The generated source module/script instance is not accessible, so we can use
-// a Map without memory concerns. Separate generated source entries with the module
-// source entries to avoid overriding the module source entries with arbitrary
-// source url magic comments.
+// The generated source module/script instance is not accessible, so these entries
+// cannot be keyed weakly like the module source entries are. The cache holds the
+// least recently used ones within a byte budget instead, or generated sources
+// evaluated under a changing source url would retain every payload they ever
+// mapped. The budget counts bytes rather than entries because one source map can
+// carry hundreds of kilobytes of sourcesContent while the next carries a handful.
+// Separate generated source entries with the module source entries to avoid
+// overriding the module source entries with arbitrary source url magic comments.
+const kGeneratedSourceMapCacheSizeLimit = 32 * 1024 * 1024;
 const generatedSourceMapCache = new SafeMap();
+let generatedSourceMapCacheSize = 0;
 const kLeadingProtocol = /^\w+:\/\//;
 const kSourceMappingURLMagicComment = /\/[*/]#\s+sourceMappingURL=(?<sourceMappingURL>[^\s]+)/g;
 const kSourceURLMagicComment = /\/[*/]#\s+sourceURL=(?<sourceURL>[^\s]+)/g;
@@ -189,7 +196,21 @@ function maybeCacheSourceMap(filename, content, moduleInstance, isGeneratedSourc
   };
 
   if (isGeneratedSource) {
+    // Only generated entries are charged against the budget, so the presence of
+    // a size doubles as the marker for an entry that is accounted for.
+    entry.size = sourceMapURL.length + entry.lineLengths.length;
+    if (entry.data != null) {
+      entry.size += sourceMapPayloadSize(entry.data);
+    }
+    deleteGeneratedSourceMap(filename);
     generatedSourceMapCache.set(filename, entry);
+    generatedSourceMapCacheSize += entry.size;
+    // Keep the newest entry even when it alone exceeds the budget, otherwise a
+    // single large source map could never be mapped at all.
+    while (generatedSourceMapCacheSize > kGeneratedSourceMapCacheSizeLimit &&
+           generatedSourceMapCache.size > 1) {
+      deleteGeneratedSourceMap(generatedSourceMapCache.keys().next().value);
+    }
     return;
   }
   // If it is not a generated source, we assume we are in a "cjs/esm"
@@ -198,6 +219,35 @@ function maybeCacheSourceMap(filename, content, moduleInstance, isGeneratedSourc
   getModuleSourceMapCache().set(keys, entry, moduleInstance);
 }
 
+/**
+ * Approximate the bytes a resolved payload keeps alive.
+ * @param {object} data - deserialized source map JSON object
+ * @returns {number} size in bytes
+ */
+function sourceMapPayloadSize(data) {
+  let size = data.mappings?.length ?? 0;
+  const sourcesContent = data.sourcesContent;
+  if (ArrayIsArray(sourcesContent)) {
+    for (let i = 0; i < sourcesContent.length; i++) {
+      size += sourcesContent[i]?.length ?? 0;
+    }
+  }
+  return size;
+}
+
+/**
+ * Drop a generated source entry and give its bytes back to the budget.
+ * @param {string} filename - key of the entry
+ */
+function deleteGeneratedSourceMap(filename) {
+  const entry = generatedSourceMapCache.get(filename);
+  if (entry === undefined) {
+    return;
+  }
+  generatedSourceMapCacheSize -= entry.size;
+  generatedSourceMapCache.delete(filename);
+}
+
 /**
  * Caches the source map if it is present in the eval'd source.
  * @param {string} content - the eval'd source code
@@ -387,6 +437,11 @@ function sourceMapCacheToObject() {
 function sourceMapData(entry) {
   if (entry.data === undefined) {
     entry.data = dataFromUrl(entry.filename, entry.sourceMapURL);
+    if (entry.size !== undefined && entry.data !== null) {
+      const size = sourceMapPayloadSize(entry.data);
+      entry.size += size;
+      generatedSourceMapCacheSize += size;
+    }
   }
   return entry.data;
 }
@@ -419,7 +474,16 @@ function findSourceMap(sourceURL) {
       // If the sourceURL is an invalid path, this will throw an error.
       sourceURL = pathToFileURL(sourceURL).href;
     }
-    const entry = getModuleSourceMapCache().get(sourceURL) ?? generatedSourceMapCache.get(sourceURL);
+    let entry = getModuleSourceMapCache().get(sourceURL);
+    if (entry === undefined) {
+      entry = generatedSourceMapCache.get(sourceURL);
+      if (entry !== undefined) {
+        // Move the entry back to the newest end, so that a generated source that
+        // is still in use is not evicted for being old.
+        generatedSourceMapCache.delete(sourceURL);
+        generatedSourceMapCache.set(sourceURL, entry);
+      }
+    }
     if (entry === undefined || sourceMapData(entry) === null) {
       return undefined;
     }
@@ -490,6 +554,7 @@ function getSourceLine(
 }
 
 module.exports = {
+  kGeneratedSourceMapCacheSizeLimit,
   findSourceMap,
   getSourceLine,
   getSourceMapsSupport,

test/parallel/test-source-map-generated-cache-limit.js

@@ -0,0 +1,49 @@
+// Flags: --enable-source-maps --expose-internals
+'use strict';
+
+/**
+ * This test verifies that the cache of source maps of generated sources stays
+ * within its budget, and that reading an entry keeps it from being evicted.
+ */
+
+require('../common');
+const assert = require('node:assert');
+const { findSourceMap } = require('node:module');
+const {
+  kGeneratedSourceMapCacheSizeLimit,
+} = require('internal/source_map/source_map_cache');
+
+const sourcesContent = 'x'.repeat(1024 * 1024);
+const payload = Buffer.from(JSON.stringify({
+  version: 3,
+  sources: ['a.js'],
+  sourcesContent: [sourcesContent],
+  names: [],
+  mappings: 'AAAA',
+})).toString('base64');
+
+function evaluate(id) {
+  eval(`(() => {})\n` +
+       `//# sourceURL=file:///generated-${id}.js\n` +
+       `//# sourceMappingURL=data:application/json;base64,${payload}`);
+}
+
+// Enough generated sources to overrun the budget twice over.
+const count = Math.ceil(kGeneratedSourceMapCacheSizeLimit / payload.length) * 2;
+
+evaluate(0);
+evaluate(1);
+assert.ok(findSourceMap('file:///generated-0.js'));
+
+for (let i = 2; i <= count; i++) {
+  evaluate(i);
+  // Reading the first entry keeps it alive while the second one ages out.
+  findSourceMap('file:///generated-0.js');
+}
+
+assert.ok(findSourceMap('file:///generated-0.js'));
+assert.strictEqual(findSourceMap('file:///generated-1.js'), undefined);
+
+const sourceMap = findSourceMap(`file:///generated-${count}.js`);
+assert.ok(sourceMap);
+assert.strictEqual(sourceMap.findEntry(0, 0).originalLine, 0);