Archive for the 'Fuzzing' Category

Some differences between JavaScript engines

Tuesday, December 23rd, 2008

I gave my new fuzzer a break from testing TraceMonkey by asking it to look for differences between SpiderMonkey and JavaScriptCore. I have listed them below, with SpiderMonkey output above JavaScriptCore output.

I have no idea how many of these are bugs (in SpiderMonkey or JavaScriptCore) and how many are ambiguous in the spec (intentionally or unintentionally).

Early error reporting

SpiderMonkey reports some errors at compile time that JavaScriptCore only reports at run time, if the code is actually hit. The difference is most obvious (and most likely to cause compatibility problems) if the code is skipped.

> if (false) { --1; }
S: SyntaxError: invalid decrement operand
J: (no error)
> if (false) { return; }
S: SyntaxError: return not in function
J: (no error)


The two engines disagree about what objects are reasonable operands for the 'instanceof' operator.

> ({} instanceof {a:2})
S: typein:3: TypeError: invalid 'instanceof' operand ({a:2})
J: false
> ({} instanceof eval)
S: false
J: Exception: TypeError: instanceof called on an object with an invalid prototype property.

new with native functions

SpiderMonkey allows the "new" operator to be used with some native functions that JavaScriptCore considers non-constructors.

> new Math.sqrt(16)
S: 4
J: Exception: TypeError: Result of expression 'Math.sqrt' ... is not a constructor.
> new ({}.toString)
S: [object Object]
J: Exception: TypeError: Result of expression '({}.toString)' ... is not a constructor.
> new eval
S: typein:9: EvalError: function eval must be called directly, and not by way of a function of another name
J: Exception: TypeError: Result of expression 'eval' ... is not a constructor.

Converting between numbers and strings

> print(+'\00000027')
S: NaN
J: 0
> (1e-10).toString(16)
S: 0.000000006df37f675ef6ec
J: 0


There are subtle differences in handling of this new keyword.

> const d; const d;
S: TypeError: redeclaration of const d
J: (no error)
> const c = 0; print(++c);
S: 0
J: 1

Other differences

> print((function(){return arguments;})());
S: [object Object]
J: [object Arguments]
> typeof /x/
S: object
J: function

See Mozilla bug 61911, which changed this in SpiderMonkey in 2007.

Fuzzing TraceMonkey

Tuesday, December 23rd, 2008

Making JavaScript faster is important for the future of computer security. Faster scripts will allow computationally intensive applications to move to the Web. As messy as the Web's security model is, it beats the most popular alternative, which is to give hundreds of native applications access to your files. Faster scripts will also allow large parts of Firefox to be written in JavaScript, a memory-safe programming language, rather than C++, a dangerous footgun.

Mozilla's ambitious TraceMonkey project adds a just-in-time compiler to Firefox's JavaScript engine, making many scripts 3 to 30 times faster. TraceMonkey takes a non-traditional approach to JIT compilation: instead of compiling a function at a time, it compiles only a path (such as the body of a loop) at a time. This makes it possible to optimize the native code based on the actual type of each variable, which is important for dynamic languages like JavaScript.

My existing JavaScript fuzzer, jsfunfuzz, found a decent number of crash and assertion bugs in early versions of TraceMonkey. I made several changes to jsfunfuzz to help it generate code to test the JIT infrastructure heavily. For example, it now generates mixed-type arrays in order to test how the JIT deals with unexpected type changes.

Andreas Gal commented that each fuzz-generated testcase saved him nearly a day of debugging: otherwise, he'd probably have to tease a testcase out of a misbehaving complex web page. Encouraged by his comment, I looked for additional ways to help the TraceMonkey team.

JIT correctness

Differential testing is designed to find correctness bugs. It runs a randomly-generated script twice (with and without the JIT) and complains if the output is different.

It quickly found 13 bugs where the JIT caused JavaScript code to produce incorrect results. These bugs range from obvious to obscure to evil.

It even found at least one security bug that jsfunfuzz had missed. An uninitialized-memory-read bug caused output to be random when it should have been consistent. jsfunfuzz missed the bug because it ignores most output, but the differential testing caught it just like it would catch a JIT vs interpreter difference.

JIT speed

I set up the new fuzzer to compare the time needed to execute scripts and complain whenever enabling the JIT made a script run more slowly. It measures speed by letting the script run for 500ms and reporting the number of loop iterations completed in that time.

So far, it has found 4 serious bugs where the JIT makes scripts several times slower. Two of these have already been fixed, but the other two may be difficult to fix.

It has also found 10 cases where the JIT makes scripts about 10% slower. Most of these minor slowdowns are due to "trace aborts", where a piece of JavaScript is not converted to native code and stays in the interpreter. Some trace aborts are due to bugs, while others are design decisions or cases for which conversion to native code simply hasn't been implemented yet.

There is some disagreement over which trace aborts are most likely to affect real web pages. I asked members of Mozilla's QA team to scan the web in a way that can answer this question.

Interpreter speed

Mostly for fun, I also looked to see which code the JIT speeds up the most. Here's a simplified version of its answer:

for (var i = 0; i < 0x02000000; ++i) {
  d = 0x55555555;
  d++; d++; d++; d++; d++;

This code runs 250 times faster when the JIT is enabled. The JIT is able to achieve this gigantic speedup due to the interpreter being inefficient in dealing with undeclared variables and numbers that can't be represented as 30-bit ints.


The JavaScript engine team has documented many of their assumptions as assertions in the code. Many of these assertions make it easier to spot dangerous bugs, because the script generated by the fuzzer doesn't have to be clever enough to actually cause a crash, only strange enough to violate an assumption. This is similar to my experience with other parts of Gecko that use assertions well.

Other JavaScript engine assertions make it easier to find severe performance bugs. Without these assertions, I'd only find these bugs when I measure speed directly, which requires drastically slowing down the tests.

More ideas

One testcase generated by my fuzzer demonstrated a combination of a JIT performance bug with a minor bytecode generation bug. I might be able to search for similar bytecode generation bugs the same way I searched for decompiler bugs: by ensuring that a function does not change when round-tripping through the decompiler. In order to do that, I'll need a new patch for making dis() return the disassembly instead of printing it.

I should be able to find some performance bugs by looking at which aborts and side exits are taken. This strategy would make some performance bugs (such as repeatedly taking a side exit) easier to spot.

Gecko assertions

Tuesday, November 27th, 2007

Assertions frequently help me find bugs that would be difficult to catch otherwise. I've filed 479 bugs on assertion failures, excluding bugs that also causes crashes or hangs, and 223 of those have already been fixed. (Thanks to Brendan, I can safely file bugs when I see assertion failures, knowing that my bugs aren't invalid.)

Many assertion failures indicate that assumptions were violated in a relatively harmless way, merely causing incorrect layout or slow performance in an edge case. But sometimes they indicate the presence of a bug that could lead to a serious memory safety violation. For example, "ASSERTION: Some objects allocated with AllocateFrame were not freed" in the FrameArena destructor has helped find dozens of bugs that lead to leaks and/or dangling pointers.

Gecko developers have added many useful assertions lately, and it looks like more are on the way. Regression tests for assertions are improving, too.

I have a list of assertions that I ignore during automated testing. Whenever a developer fixes a bug on that list, I remove the bug and the corresponding assertion from the file, so if I hit the assertion again after that point, I'll notice it.

In addition to helping to catch bugs, assertions also serve as "living documentation" about the code's invariants. We notice if they become out of date, unlike Wiki pages and comments. In my dream world, assertions would also serve as waypoints for an automated theorem prover ;)

Valgrind coming to Mac

Wednesday, September 19th, 2007

Apple employee Greg Parker has ported Valgrind to Mac, and plans to release his work soon after Leopard is released in October. He's been working on it for quite a while.

I'm excited about being able to use Valgrind on Mac. Valgrind's "Memcheck" is much better at catching dangling-pointer bugs and heap buffer-overflow bugs than simply watching for crashes (even with MallocScribble enabled). Running a fuzzer with Memcheck can reveal exploitable memory safety bugs that would not have triggered crashes otherwise.

Update 2008-09-29: Greg Parker has released his port as a patch.

Update 2009-02-06: Valgrind developer Nicholas Nethercote has imported Greg's patch into a branch of Valgrind's SVN repository.

Update 2009-06-02: Valgrind trunk now supports Mac.

Introducing Lithium, a testcase reduction tool

Saturday, September 15th, 2007

I wrote a tool called Lithium that automatically reduces large testcases, such as real-world web pages or testcases produced by jsfunfuzz. It can usually reduce a 3000-line jsfunfuzz crash testcase to 3-10 lines in several minutes, considerably faster than I can reduce by hand. Perhaps more importantly, I can do something else while it reduces the testcase.

There are two (related) reasons I'm not calling it "Lithium 1.0" yet. First, I'm hoping to improve the way "interestingness tests" are written. Currently, they're separate programs that communicate to Lithium using their exit code, which limits error handling and might slow Lithium down. I'd like to make the interestingness tests be Python files, but I'm not sure what the best way to do that is. (Should Lithium __import__ the interestingness test? Or should the interestingness test import Lithium and be renamed to e.g. ""?)

Second, it would be useful to be able to pass extra arguments to the program being tested. For example, it would be useful to be able to pass a profile name to Firefox, or to pass a Firefox path to Valgrind. One possibility is to put the program being tested last on the command line, so extra positional arguments become options to that program. This solution would only work for interestingness tests that launch a single program (so it wouldn't work for a "renders differently in these two Firefox builds" test, for example), but maybe that's okay. Another possibility is to require the use of a config file for passing arguments to programs being tested (so you don't end up typing all of ".../firefox-bin -P foo" on Lithium's command line).

I'll probably use the MIT license for Lithium (but not for, which was mostly written by Chris Cooper and Bob Clary).

Opera is finding jsfunfuzz useful

Saturday, August 4th, 2007

Opera has posted a build with fixes for several crashes found by jsfunfuzz. Cool!

Opera community members posted dozens of comments about it, and I replied to several.

jsfunfuzz in news and blogs

Friday, August 3rd, 2007

Before the presentation:

After the presentation:

Fuzzing for JavaScript correctness

Thursday, August 2nd, 2007

Fuzz-testing is usually only used to find crashes and assertion failures, but my JavaScript engine fuzzer goes beyond catastrophic failures when it tests the decompiler. It checks the decompiled code for signs of incorrectness in two ways.

First, it checks that the decompiled code compiles without giving syntax errors. This finds fun bugs like bug 346904 where the decompiler screwed up in an understandable way, as well as bugs like bug 351496 where the decompilation is complete nonsense.

Second, it checks that the decompiled code is canonical -- compiling and decompiling again should give the exact same representation as the original decompilation. This helps find bugs like bug 381196 where decompilation changes the meaning of the code without introducing a syntax error.

Some decompilation changes, such as bug 352068, did not change the meaning of the code and simply reflected varying amounts of optimization in the compiler. Early in the fuzzer's life, I was able to convince Brendan that it was worth fixing many of those otherwise harmless "round-trip changes" in order to make it possible to find other bugs with this method.

This pair of checks doesn't find all decompiler bugs, of course, but it finds quite a few of them. jsfunfuzz has a few other correctness checks for things like unnecessary parentheses in decompiled code and bogus results from object uneval.

Can you think of other ways to use fuzz-testing to find "correctness" bugs?