Merge pull request #132 from zeux/fuzz

Improve fuzzing support

Makefile

@@ -68,10 +68,9 @@ test: $(EXECUTABLE)
 	./$(EXECUTABLE)
 endif
 
-fuzz:
-	@mkdir -p $(BUILD)
-	$(AFL)/afl-clang++ tests/fuzz_parse.cpp tests/allocator.cpp src/pugixml.cpp $(CXXFLAGS) -o $(BUILD)/fuzz_parse
-	$(AFL)/afl-fuzz -i tests/data_fuzz_parse -o $(BUILD)/fuzz_parse_out -x $(AFL)/testcases/_extras/xml/ -- $(BUILD)/fuzz_parse @@
+fuzz_%: $(BUILD)/fuzz_%
+	@mkdir -p build/$@
+	$< build/$@ tests/data_$* -max_len=1024 -dict=tests/fuzz_$*.dict
 
 clean:
 	rm -rf $(BUILD)
@@ -87,6 +86,10 @@ build/pugixml-%: .FORCE | $(RELEASE)
 $(EXECUTABLE): $(OBJECTS)
 	$(CXX) $(OBJECTS) $(LDFLAGS) -o $@
 
+$(BUILD)/fuzz_%: tests/fuzz_%.cpp src/pugixml.cpp
+	@mkdir -p $(BUILD)
+	clang++ $(CXXFLAGS) -fsanitize=address -fsanitize-coverage=trace-pc-guard $^ libFuzzer.a -o $@
+
 $(BUILD)/%.o: %
 	@mkdir -p $(dir $@)
 	$(CXX) $< $(CXXFLAGS) -c -MMD -MP -o $@

tests/data_fuzz_xpath/basic.xpath

@@ -0,0 +1 @@
+a/b/c

tests/data_fuzz_xpath/functions.xpath

@@ -0,0 +1 @@
+sum(nodes) + round(concat(//a[translate(@id, 'abc', '012')]))

tests/data_fuzz_xpath/math.xpath

@@ -0,0 +1 @@
+1+2*3 div 4 mod 5-6

tests/data_fuzz_xpath/path.xpath

@@ -0,0 +1 @@
+@*/ancestor::*/near-north/*[4]/@*/preceding::text()

tests/data_fuzz_xpath/predicate.xpath

@@ -0,0 +1 @@
+library/nodes[@id=12]/element[@type='translate'][1]

tests/fuzz_parse.cpp

@@ -1,16 +1,14 @@
 #include "../src/pugixml.hpp"
-#include "allocator.hpp"
 
-int main(int argc, const char** argv)
+#include <stdint.h>
+
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t *Data, size_t Size)
 {
-    pugi::set_memory_management_functions(memory_allocate, memory_deallocate);
-
     pugi::xml_document doc;
 
-    for (int i = 1; i < argc; ++i)
-    {
-	    doc.load_file(argv[i]);
-	    doc.load_file(argv[i], pugi::parse_minimal);
-	    doc.load_file(argv[i], pugi::parse_full);
-	}
+	doc.load_buffer(Data, Size);
+	doc.load_buffer(Data, Size, pugi::parse_minimal);
+	doc.load_buffer(Data, Size, pugi::parse_full);
+
+	return 0;
 }

tests/fuzz_parse.dict

@@ -0,0 +1,72 @@
+#
+# AFL dictionary for XML
+# ----------------------
+#
+# Several basic syntax elements and attributes, modeled on libxml2.
+#
+# Created by Michal Zalewski <lcamtuf@google.com>
+#
+
+attr_encoding=" encoding=\"1\""
+attr_generic=" a=\"1\""
+attr_href=" href=\"1\""
+attr_standalone=" standalone=\"no\""
+attr_version=" version=\"1\""
+attr_xml_base=" xml:base=\"1\""
+attr_xml_id=" xml:id=\"1\""
+attr_xml_lang=" xml:lang=\"1\""
+attr_xml_space=" xml:space=\"1\""
+attr_xmlns=" xmlns=\"1\""
+
+entity_builtin="&lt;"
+entity_decimal="&#1;"
+entity_external="&a;"
+entity_hex="&#x1;"
+
+string_any="ANY"
+string_brackets="[]"
+string_cdata="CDATA"
+string_col_fallback=":fallback"
+string_col_generic=":a"
+string_col_include=":include"
+string_dashes="--"
+string_empty="EMPTY"
+string_empty_dblquotes="\"\""
+string_empty_quotes="''"
+string_entities="ENTITIES"
+string_entity="ENTITY"
+string_fixed="#FIXED"
+string_id="ID"
+string_idref="IDREF"
+string_idrefs="IDREFS"
+string_implied="#IMPLIED"
+string_nmtoken="NMTOKEN"
+string_nmtokens="NMTOKENS"
+string_notation="NOTATION"
+string_parentheses="()"
+string_pcdata="#PCDATA"
+string_percent="%a"
+string_public="PUBLIC"
+string_required="#REQUIRED"
+string_schema=":schema"
+string_system="SYSTEM"
+string_ucs4="UCS-4"
+string_utf16="UTF-16"
+string_utf8="UTF-8"
+string_xmlns="xmlns:"
+
+tag_attlist="<!ATTLIST"
+tag_cdata="<![CDATA["
+tag_close="</a>"
+tag_doctype="<!DOCTYPE"
+tag_element="<!ELEMENT"
+tag_entity="<!ENTITY"
+tag_ignore="<![IGNORE["
+tag_include="<![INCLUDE["
+tag_notation="<!NOTATION"
+tag_open="<a>"
+tag_open_close="<a />"
+tag_open_exclamation="<!"
+tag_open_q="<?"
+tag_sq2_close="]]>"
+tag_xml_q="<?xml?>"

tests/fuzz_setup.sh

@@ -0,0 +1,13 @@
+#!/bin/bash
+
+sudo apt-get --yes install subversion screen gcc g++ cmake ninja-build golang autoconf libtool apache2 python-dev pkg-config zlib1g-dev libgcrypt11-dev
+
+cd clang
+git clone https://chromium.googlesource.com/chromium/src/tools/clang
+cd ..
+clang/clang/scripts/update.py
+sudo cp -rf third_party/llvm-build/Release+Asserts/lib/* /usr/local/lib/
+sudo cp -rf third_party/llvm-build/Release+Asserts/bin/* /usr/local/bin
+
+svn co http://llvm.org/svn/llvm-project/llvm/trunk/lib/Fuzzer
+Fuzzer/build.sh

tests/fuzz_xpath.cpp

@@ -0,0 +1,26 @@
+#include "../src/pugixml.hpp"
+
+#include <stdint.h>
+#include <string.h>
+
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t *Data, size_t Size)
+{
+	char* text = new char[Size + 1];
+	memcpy(text, Data, Size);
+	text[Size] = 0;
+
+#ifdef PUGIXML_NO_EXCEPTIONS
+	pugi::xpath_query q(text);
+#else
+	try
+	{
+		pugi::xpath_query q(text);
+	}
+	catch (pugi::xpath_exception&)
+	{
+	}
+#endif
+
+	delete[] text;
+	return 0;
+}

tests/fuzz_xpath.dict

@@ -0,0 +1,72 @@
+"boolean"
+"count"
+"contains"
+"concat"
+"ceiling"
+"false"
+"floor"
+"id"
+"last"
+"lang"
+"local-name"
+"name"
+"namespace-uri"
+"normalize-space"
+"not"
+"number"
+"position"
+"round"
+"string"
+"string-length"
+"starts-with"
+"substring-before"
+"substring-after"
+"substring"
+"sum"
+"translate"
+"true"
+"ancestor"
+"ancestor-or-self"
+"attribute"
+"child"
+"descendant"
+"descendant-or-self"
+"following"
+"following-sibling"
+"namespace"
+"parent"
+"preceding"
+"preceding-sibling"
+"self"
+"comment"
+"node"
+"processing-instruction"
+"text"
+"or"
+"and"
+"div"
+"mod"
+">"
+">="
+"<"
+"<="
+"!"
+"!="
+"="
+"+"
+"-"
+"*"
+"|"
+"$"
+"("
+")"
+"["
+"]"
+","
+"//"
+"/"
+".."
+"."
+"@"
+"::"
+":"