Bug 561874 - Make the HTML5 parser clip deep trees (similar to the old HTML parser) to avoid crashes in recursive code elsewhere. r=bzbarsky

parser/html/javasrc/TreeBuilder.java

@@ -59,8 +59,6 @@ import org.xml.sax.SAXParseException;
 
 public abstract class TreeBuilder<T> implements TokenHandler,
         TreeBuilderState<T> {
-
-    public static final int STACK_MAX_DEPTH = 200;
     
     // Start dispatch groups
 
@@ -4217,10 +4215,6 @@ public abstract class TreeBuilder<T> implements TokenHandler,
     }
 
     @SuppressWarnings("unchecked") private void push(StackNode<T> node) throws SAXException {
-        if (currentPtr == TreeBuilder.STACK_MAX_DEPTH) {
-            warn("Maximum depth for tree builder stack reached. Modifying document.");
-            pop();
-        }
         currentPtr++;
         if (currentPtr == stack.length) {
             StackNode<T>[] newStack = new StackNode[stack.length + 64];
@@ -4233,10 +4227,6 @@ public abstract class TreeBuilder<T> implements TokenHandler,
     }
 
     @SuppressWarnings("unchecked") private void silentPush(StackNode<T> node) throws SAXException {
-        if (currentPtr == TreeBuilder.STACK_MAX_DEPTH) {
-            warn("Maximum depth for tree builder stack reached. Modifying document.");
-            pop();
-        }
         currentPtr++;
         if (currentPtr == stack.length) {
             StackNode<T>[] newStack = new StackNode[stack.length + 64];

parser/html/nsHtml5TreeBuilder.cpp

@@ -3013,10 +3013,6 @@ nsHtml5TreeBuilder::clearLastListSlot()
 void 
 nsHtml5TreeBuilder::push(nsHtml5StackNode* node)
 {
-  if (currentPtr == NS_HTML5TREE_BUILDER_STACK_MAX_DEPTH) {
-
-    pop();
-  }
   currentPtr++;
   if (currentPtr == stack.length) {
     jArray<nsHtml5StackNode*,PRInt32> newStack = jArray<nsHtml5StackNode*,PRInt32>(stack.length + 64);
@@ -3031,10 +3027,6 @@ nsHtml5TreeBuilder::push(nsHtml5StackNode* node)
 void 
 nsHtml5TreeBuilder::silentPush(nsHtml5StackNode* node)
 {
-  if (currentPtr == NS_HTML5TREE_BUILDER_STACK_MAX_DEPTH) {
-
-    pop();
-  }
   currentPtr++;
   if (currentPtr == stack.length) {
     jArray<nsHtml5StackNode*,PRInt32> newStack = jArray<nsHtml5StackNode*,PRInt32>(stack.length + 64);

parser/html/nsHtml5TreeBuilder.h

@@ -242,7 +242,6 @@ class nsHtml5TreeBuilder : public nsAHtml5TreeBuilderState
 jArray<const char*,PRInt32> nsHtml5TreeBuilder::QUIRKY_PUBLIC_IDS = nsnull;
 #endif
 
-#define NS_HTML5TREE_BUILDER_STACK_MAX_DEPTH 200
 #define NS_HTML5TREE_BUILDER_OTHER 0
 #define NS_HTML5TREE_BUILDER_A 1
 #define NS_HTML5TREE_BUILDER_BASE 2

parser/html/nsHtml5TreeBuilderCppSupplement.h

@@ -46,6 +46,8 @@
 #include "nsContentUtils.h"
 #include "nsNodeUtils.h"
 
+#define NS_HTML5_TREE_DEPTH_LIMIT 200
+
 class nsPresContext;
 
 // this really should be autogenerated...
@@ -243,7 +245,9 @@ nsHtml5TreeBuilder::appendElement(nsIContent** aChild, nsIContent** aParent)
 {
   NS_PRECONDITION(aChild, "Null child");
   NS_PRECONDITION(aParent, "Null parent");
-
+  if (mDeepTreeSurrogateParent) {
+    return;
+  }
   nsHtml5TreeOperation* treeOp = mOpQueue.AppendElement();
   NS_ASSERTION(treeOp, "Tree op allocation failed.");
   treeOp->Init(eTreeOpAppend, aChild, aParent);
@@ -298,7 +302,8 @@ nsHtml5TreeBuilder::appendCharacters(nsIContent** aParent, PRUnichar* aBuffer, P
   
   nsHtml5TreeOperation* treeOp = mOpQueue.AppendElement();
   NS_ASSERTION(treeOp, "Tree op allocation failed.");
-  treeOp->Init(eTreeOpAppendText, bufferCopy, aLength, aParent);
+  treeOp->Init(eTreeOpAppendText, bufferCopy, aLength,
+      mDeepTreeSurrogateParent ? mDeepTreeSurrogateParent : aParent);
 }
 
 void
@@ -306,6 +311,9 @@ nsHtml5TreeBuilder::appendComment(nsIContent** aParent, PRUnichar* aBuffer, PRIn
 {
   NS_PRECONDITION(aBuffer, "Null buffer");
   NS_PRECONDITION(aParent, "Null parent");
+  if (mDeepTreeSurrogateParent) {
+    return;
+  }
 
   PRUnichar* bufferCopy = new PRUnichar[aLength];
   memcpy(bufferCopy, aBuffer, aLength * sizeof(PRUnichar));
@@ -356,6 +364,7 @@ void
 nsHtml5TreeBuilder::start(PRBool fragment)
 {
   mCurrentHtmlScriptIsAsyncOrDefer = PR_FALSE;
+  mDeepTreeSurrogateParent = nsnull;
 #ifdef DEBUG
   mActive = PR_TRUE;
 #endif
@@ -388,6 +397,36 @@ nsHtml5TreeBuilder::elementPushed(PRInt32 aNamespace, nsIAtom* aName, nsIContent
   NS_ASSERTION(aNamespace == kNameSpaceID_XHTML || aNamespace == kNameSpaceID_SVG || aNamespace == kNameSpaceID_MathML, "Element isn't HTML, SVG or MathML!");
   NS_ASSERTION(aName, "Element doesn't have local name!");
   NS_ASSERTION(aElement, "No element!");
+  /*
+   * The frame constructor uses recursive algorithms, so it can't deal with
+   * arbitrarily deep trees. This is especially a problem on Windows where
+   * the permitted depth of the runtime stack is rather small.
+   *
+   * The following is a protection against author incompetence--not against
+   * malice. There are other ways to make the DOM deep anyway.
+   *
+   * The basic idea is that when the tree builder stack gets too deep,
+   * append operations no longer append to the node that the HTML parsing
+   * algorithm says they should but instead text nodes are append to the last
+   * element that was seen before a magic tree builder stack threshold was
+   * reached and element and comment nodes aren't appended to the DOM at all.
+   *
+   * However, for security reasons, non-child descendant text nodes inside an
+   * SVG script or style element should not become children. Also, non-cell
+   * table elements shouldn't be used as surrogate parents for user experience
+   * reasons.
+   */
+  if (!mDeepTreeSurrogateParent && currentPtr >= NS_HTML5_TREE_DEPTH_LIMIT &&
+      !(aName == nsHtml5Atoms::script ||
+        aName == nsHtml5Atoms::table ||
+        aName == nsHtml5Atoms::thead ||
+        aName == nsHtml5Atoms::tfoot ||
+        aName == nsHtml5Atoms::tbody ||
+        aName == nsHtml5Atoms::tr ||
+        aName == nsHtml5Atoms::colgroup ||
+        aName == nsHtml5Atoms::style)) {
+    mDeepTreeSurrogateParent = aElement;
+  }
   if (aNamespace != kNameSpaceID_XHTML) {
     return;
   }
@@ -405,6 +444,9 @@ nsHtml5TreeBuilder::elementPopped(PRInt32 aNamespace, nsIAtom* aName, nsIContent
   NS_ASSERTION(aNamespace == kNameSpaceID_XHTML || aNamespace == kNameSpaceID_SVG || aNamespace == kNameSpaceID_MathML, "Element isn't HTML, SVG or MathML!");
   NS_ASSERTION(aName, "Element doesn't have local name!");
   NS_ASSERTION(aElement, "No element!");
+  if (mDeepTreeSurrogateParent && currentPtr <= NS_HTML5_TREE_DEPTH_LIMIT) {
+    mDeepTreeSurrogateParent = nsnull;
+  }
   if (aNamespace == kNameSpaceID_MathML) {
     return;
   }

parser/html/nsHtml5TreeBuilderHSupplement.h

@@ -46,6 +46,7 @@
     PRInt32                                mHandlesUsed;
     nsTArray<nsAutoArrayPtr<nsIContent*> > mOldHandles;
     nsHtml5TreeOpStage*                    mSpeculativeLoadStage;
+    nsIContent**                           mDeepTreeSurrogateParent;
     PRBool                                 mCurrentHtmlScriptIsAsyncOrDefer;
 #ifdef DEBUG
     PRBool                                 mActive;