Avoid overdeep recursion in regexp where a guarded expression with a

src/jsregexp.cc

Index: src/jsregexp.cc
===================================================================
--- src/jsregexp.cc	(revision 11684)
+++ src/jsregexp.cc	(working copy)
@@ -2191,12 +2191,13 @@
 
 
 void ActionNode::FillInBMInfo(int offset,
+                              int recursion_depth,
                               BoyerMooreLookahead* bm,
                               bool not_at_start) {
   if (type_ == BEGIN_SUBMATCH) {
     bm->SetRest(offset);
   } else if (type_ != POSITIVE_SUBMATCH_SUCCESS) {
-    on_success()->FillInBMInfo(offset, bm, not_at_start);
+    on_success()->FillInBMInfo(offset, recursion_depth + 1, bm, not_at_start);
   }
   SaveBMInfo(bm, not_at_start, offset);
 }
@@ -2218,11 +2219,13 @@
 }
 
 
-void AssertionNode::FillInBMInfo(
-    int offset, BoyerMooreLookahead* bm, bool not_at_start) {
+void AssertionNode::FillInBMInfo(int offset,
+                                 int recursion_depth,
+                                 BoyerMooreLookahead* bm,
+                                 bool not_at_start) {
   // Match the behaviour of EatsAtLeast on this node.
   if (type() == AT_START && not_at_start) return;
-  on_success()->FillInBMInfo(offset, bm, not_at_start);
+  on_success()->FillInBMInfo(offset, recursion_depth + 1, bm, not_at_start);
   SaveBMInfo(bm, not_at_start, offset);
 }
 
@@ -2803,14 +2806,17 @@
 }
 
 
-void LoopChoiceNode::FillInBMInfo(
-    int offset, BoyerMooreLookahead* bm, bool not_at_start) {
-  if (body_can_be_zero_length_) {
+void LoopChoiceNode::FillInBMInfo(int offset,
+                                  int recursion_depth,
+                                  BoyerMooreLookahead* bm,
+                                  bool not_at_start) {
+  if (body_can_be_zero_length_ ||
+      recursion_depth > RegExpCompiler::kMaxRecursion) {
     bm->SetRest(offset);
     SaveBMInfo(bm, not_at_start, offset);
     return;
   }
-  ChoiceNode::FillInBMInfo(offset, bm, not_at_start);
+  ChoiceNode::FillInBMInfo(offset, recursion_depth + 1, bm, not_at_start);
   SaveBMInfo(bm, not_at_start, offset);
 }
 
@@ -2912,7 +2918,7 @@
     if (eats_at_least >= 1) {
       BoyerMooreLookahead* bm =
           new BoyerMooreLookahead(eats_at_least, compiler);
-      FillInBMInfo(0, bm, not_at_start);
+      FillInBMInfo(0, 0, bm, not_at_start);
       if (bm->at(0)->is_non_word()) next_is_word_character = Trace::FALSE;
       if (bm->at(0)->is_word()) next_is_word_character = Trace::TRUE;
     }
@@ -3850,7 +3856,7 @@
             BoyerMooreLookahead* bm =
                 new BoyerMooreLookahead(eats_at_least, compiler);
             GuardedAlternative alt0 = alternatives_->at(0);
-            alt0.node()->FillInBMInfo(0, bm, not_at_start);
+            alt0.node()->FillInBMInfo(0, 0, bm, not_at_start);
             skip_was_emitted = bm->EmitSkipInstructions(macro_assembler);
           }
         } else {
@@ -5589,8 +5595,10 @@
 }
 
 
-void BackReferenceNode::FillInBMInfo(
-    int offset, BoyerMooreLookahead* bm, bool not_at_start) {
+void BackReferenceNode::FillInBMInfo(int offset,
+                                     int recursion_depth,
+                                     BoyerMooreLookahead* bm,
+                                     bool not_at_start) {
   // Working out the set of characters that a backreference can match is too
   // hard, so we just say that any character can match.
   bm->SetRest(offset);
@@ -5602,8 +5610,10 @@
               RegExpMacroAssembler::kTableSize);
 
 
-void ChoiceNode::FillInBMInfo(
-    int offset, BoyerMooreLookahead* bm, bool not_at_start) {
+void ChoiceNode::FillInBMInfo(int offset,
+                              int recursion_depth,
+                              BoyerMooreLookahead* bm,
+                              bool not_at_start) {
   ZoneList<GuardedAlternative>* alts = alternatives();

[ulan, 2012/05/31 11:56:23]

We don't check recursion_depth here, because nesting of choice nodes is limited
somewhere else?

[Erik Corry, 2012/05/31 13:51:43]

Yes, it's only LoopChoiceNodes that can lead to unrestricted recursion.

   for (int i = 0; i < alts->length(); i++) {
     GuardedAlternative& alt = alts->at(i);
@@ -5612,14 +5622,16 @@
       SaveBMInfo(bm, not_at_start, offset);
       return;
     }
-    alt.node()->FillInBMInfo(offset, bm, not_at_start);
+    alt.node()->FillInBMInfo(offset, recursion_depth + 1, bm, not_at_start);
   }
   SaveBMInfo(bm, not_at_start, offset);
 }
 
 
-void TextNode::FillInBMInfo(
-    int initial_offset, BoyerMooreLookahead* bm, bool not_at_start) {
+void TextNode::FillInBMInfo(int initial_offset,
+                            int recursion_depth,
+                            BoyerMooreLookahead* bm,
+                            bool not_at_start) {
   if (initial_offset >= bm->length()) return;
   int offset = initial_offset;
   int max_char = bm->max_char();
@@ -5673,6 +5685,7 @@
     return;
   }
   on_success()->FillInBMInfo(offset,
+                             recursion_depth + 1,
                              bm,
                              true);  // Not at start after a text node.
   if (initial_offset == 0) set_bm_info(not_at_start, bm);