[html parser] Check for integer overflow when computing new buffer sizes.

parser/html/java/htmlparser/src/nu/validator/htmlparser/impl/MetaScanner.java

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2007 Henri Sivonen
  * Copyright (c) 2008-2015 Mozilla Foundation
- * Copyright (c) 2018-2020 Moonchild Productions
+ * Copyright (c) 2018-2021 Moonchild Productions
  * Copyright (c) 2020 Binary Outcast
  *
  * Permission is hereby granted, free of charge, to any person obtaining a 
@@ -753,7 +753,7 @@ public abstract class MetaScanner {
         stateSave  = state;
     }
 
-    private void handleCharInAttributeValue(int c) {
+    private void handleCharInAttributeValue(int c) throws SAXException {
         if (metaState == A) {
             if (contentIndex == CONTENT.length || charsetIndex == CHARSET.length) {
                 addToBuffer(c);
@@ -778,9 +778,9 @@ public abstract class MetaScanner {
      * Adds a character to the accumulation buffer.
      * @param c the character to add
      */
-    private void addToBuffer(int c) {
+    private void addToBuffer(int c) throws SAXException {
         if (strBufLen == strBuf.length) {
-            char[] newBuf = new char[strBuf.length + (strBuf.length << 1)];
+            char[] newBuf = new char[Portability.checkedAdd(strBuf.length, (strBuf.length << 1))];
             System.arraycopy(strBuf, 0, newBuf, 0, strBuf.length);
             strBuf = newBuf;
         }

parser/html/java/htmlparser/src/nu/validator/htmlparser/impl/Portability.java

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2008-2015 Mozilla Foundation
- * Copyright (c) 2018-2020 Moonchild Productions
+ * Copyright (c) 2018-2021 Moonchild Productions
  * Copyright (c) 2020 Binary Outcast
  *
  * Permission is hereby granted, free of charge, to any person obtaining a 
@@ -24,6 +24,8 @@
 
 package nu.validator.htmlparser.impl;
 
+import org.xml.sax.SAXException;
+
 import nu.validator.htmlparser.annotation.Literal;
 import nu.validator.htmlparser.annotation.Local;
 import nu.validator.htmlparser.annotation.NoLength;
@@ -31,6 +33,17 @@ import nu.validator.htmlparser.common.Interner;
 
 public final class Portability {
 
+    public static int checkedAdd(int a, int b) throws SAXException {
+        // This can't be translated code, because in C++ signed integer overflow is UB, so the below code would be wrong.
+        assert a >= 0;
+        assert b >= 0;
+        int sum = a + b;
+        if (sum < a || sum < b) {
+            throw new SAXException("Integer overflow");
+        }
+        return sum;
+    }
+
     // Allocating methods
 
     /**

parser/html/java/htmlparser/src/nu/validator/htmlparser/impl/Tokenizer.java

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2005-2007 Henri Sivonen
  * Copyright (c) 2007-2015 Mozilla Foundation
- * Copyright (c) 2018-2020 Moonchild Productions
+ * Copyright (c) 2018-2021 Moonchild Productions
  * Copyright (c) 2020 Binary Outcast
  * Portions of comments Copyright 2004-2010 Apple Computer, Inc., Mozilla
  * Foundation, and Opera Software ASA.
@@ -1009,8 +1009,8 @@ public class Tokenizer implements Locator {
         // ]NOCPP]
     }
 
-    private void appendStrBuf(@NoLength char[] buffer, int offset, int length) {
+    private void appendStrBuf(@NoLength char[] buffer, int offset, int length) throws SAXException {
-        int newLen = strBufLen + length;
+        int newLen = Portability.checkedAdd(strBufLen, length);
         // CPPONLY: assert newLen <= strBuf.length: "Previous buffer length insufficient.";
         // CPPONLY: if (strBuf.length < newLen) {
         // CPPONLY:     if (!EnsureBufferSpace(length)) {
@@ -1024,7 +1024,7 @@ public class Tokenizer implements Locator {
     /**
      * Append the contents of the char reference buffer to the main one.
      */
-    @Inline private void appendCharRefBufToStrBuf() {
+    @Inline private void appendCharRefBufToStrBuf() throws SAXException {
         appendStrBuf(charRefBuf, 0, charRefBufLen);
         charRefBufLen = 0;
     }

parser/html/nsHtml5MetaScanner.cpp

@@ -743,7 +743,7 @@ void
 nsHtml5MetaScanner::addToBuffer(int32_t c)
 {
   if (strBufLen == strBuf.length) {
-    jArray<char16_t,int32_t> newBuf = jArray<char16_t,int32_t>::newJArray(strBuf.length + (strBuf.length << 1));
+    jArray<char16_t,int32_t> newBuf = jArray<char16_t,int32_t>::newJArray(nsHtml5Portability::checkedAdd(strBuf.length, (strBuf.length << 1)));
     nsHtml5ArrayCopy::arraycopy(strBuf, newBuf, strBuf.length);
     strBuf = newBuf;
   }

parser/html/nsHtml5Portability.cpp

@@ -7,6 +7,15 @@
 #include "jArray.h"
 #include "nsHtml5Portability.h"
 #include "nsHtml5TreeBuilder.h"
+#include "mozilla/CheckedInt.h"
+
+int32_t nsHtml5Portability::checkedAdd(int32_t a, int32_t b) {
+  mozilla::CheckedInt<int32_t> sum(a);
+  sum += b;
+  MOZ_RELEASE_ASSERT(sum.isValid(),
+                     "HTML input too large for signed 32-bit integer.");
+  return sum.value();
+}
 
 nsIAtom*
 nsHtml5Portability::newLocalNameFromBuffer(char16_t* buf, int32_t offset, int32_t length, nsHtml5AtomTable* interner)

parser/html/nsHtml5Portability.h

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2008-2015 Mozilla Foundation
- * Copyright (c) 2018-2020 Moonchild Productions
+ * Copyright (c) 2018-2021 Moonchild Productions
  * Copyright (c) 2020 Binary Outcast
  *
  * Permission is hereby granted, free of charge, to any person obtaining a 
@@ -60,6 +60,7 @@ class nsHtml5StateSnapshot;
 class nsHtml5Portability
 {
   public:
+    static int32_t checkedAdd(int32_t a, int32_t b);
     static nsIAtom* newLocalNameFromBuffer(char16_t* buf, int32_t offset, int32_t length, nsHtml5AtomTable* interner);
     static nsHtml5String newStringFromBuffer(char16_t* buf, int32_t offset, int32_t length, nsHtml5TreeBuilder* treeBuilder, bool maybeAtomize);
     static nsHtml5String newEmptyString();

parser/html/nsHtml5Tokenizer.cpp

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2005-2007 Henri Sivonen
  * Copyright (c) 2007-2015 Mozilla Foundation
- * Copyright (c) 2018-2020 Moonchild Productions
+ * Copyright (c) 2018-2021 Moonchild Productions
  * Copyright (c) 2020 Binary Outcast
  * Portions of comments Copyright 2004-2010 Apple Computer, Inc., Mozilla
  * Foundation, and Opera Software ASA.
@@ -249,7 +249,7 @@ nsHtml5Tokenizer::emitStrBuf()
 void 
 nsHtml5Tokenizer::appendStrBuf(char16_t* buffer, int32_t offset, int32_t length)
 {
-  int32_t newLen = strBufLen + length;
+  int32_t newLen = nsHtml5Portability::checkedAdd(strBufLen, length);
   MOZ_ASSERT(newLen <= strBuf.length, "Previous buffer length insufficient.");
   if (MOZ_UNLIKELY(strBuf.length < newLen)) {
     if (MOZ_UNLIKELY(!EnsureBufferSpace(length))) {