Disabling JIT in Safari 6 to workaround severe Jav

We found a severe problem with the interpretation of our Javascript code that only occurs on iOS 5/Safari 6 (then current iPad release) that we think is due to critical bug in the Just in Time JS compiler in Safari. (See updates below for more affected versions and versions that seem to now contain a fix).

We originally found the issue in our online demos of our library: the demos crash more or less randomly but this happens only the second time (or even later) that the same code is executed. I.e. if you run the part of the code once, everything works OK, however subsequent runs crash the application.

Interestingly executing the same code in Chrome for iOS the problem does not show, which we believe is due to the missing JIT capabilities of the Webview that is used in Chrome for iOS.

After a lot of fiddling we finally think we found at least one problematic piece of code:

  var a = 0; // counter for index
  for (var b = this.getStart(); b !== null; b = b.getNext()) // iterate over all cells
    b.$f = a++; // assign index to cell and then increment

In essence this is a simple for loop that assigns each cell in a linked list data structure its index. The problem here is the post-increment operation in the loop body. The current count is assigned to the field and updated after the expression is evaluated, basically the same as first assigning a and then incrementing it by one.

This works OK in all browsers we tested and in Safari for the first couple of times, and then suddenly it seems as if the counter variable a is incremented first and then the result is assigned, like a pre-increment operation.

I have created a fiddle that shows the problem here: http://jsfiddle.net/yGuy/L6t5G/

Running the example on an iPad 2 with iOS 6 and all updates the result is OK for the first 2 runs in my case and in the third identic run suddenly the last element in the list has a value assigned that is off by one (the output when you click the "click me" button changes from "from 0 to 500" to "from 0 to 501")

Interestingly if you switch tabs, or wait a little it can happen that suddenly the results are correct for two or so more runs! It seems as if Safari sometimes resets is JIT caches.

So since I think it may take a very long for the Safari team to fix this bug (which I have not yet reported) and there may be other similar bugs like this lurking in the JIT that are equally hard to find, I would like to know whether there is a way to disable the JIT functionality in Safari. Of course this would slow down our code (which is very CPU intensive already), but better slow than crashing.

Update: Unsurprisingly it's not just the post increment operator that is affected, but also the post decrement operator. Less surprisingly and more worryingly is that it makes no difference if the value is assigned, so looking for an assignment in existing code is not enough. E.g. the following the code b.$f = (a++ % 2 == 0) ? 1 : 2; where the variables value is not assigned but just used for the ternary operator condition also "fails" in the sense that sometimes the wrong branch is chosen. Currently it looks as if the problem can only be avoided if the post operators are not used at all.

Update: The same issue does not only exist in iOS devices, but also on Mac OSX in Safari 6 and the latest Safari 5: These have been tested and found to be affected by the bug: Mac OS 10.7.4, Safari 5.1.7 Mac OS X 10.8.2, WebKit Nightly r132968: Safari 6.0.1 (8536.26.14, 537+). Interestingly these do not seem to be affected: iPad 2 (Mobile) Safari 5.1.7, and iPad 1 Mobile Safari 5.1. I have reported these problems to Apple but have not received any response, yet.

Update: The bug has been reported as Webkit bug 109036. Apple still has not responded to my bug report, all current (February 2013) Safari versions on iOS and MacOS are still affected by the problem.

Update 27th of February 2013: It seems the bug has been fixed by the Webkit team here! It was indeed a problem with the JIT and the post-operators! The comments indicate that more code might have been affected by the bug, so it could be that more mysterious Heisenbugs have been fixed, now!

Update October 2013: The fix finally made it into production code: iOS 7.0.2 at least on iPad2 does not seem to suffer from this bug anymore. I did not check all of the intermediate versions, though, since we worked around the problem a long time ago.

回答1:

Try-catch blocks seem to disable the JIT compiler on Safari 6 on Lion for the part directly inside the try block (this code worked for me on Safari 6.0.1 7536.26.14 and OS X Lion).

// test function
utility.test = function(){
    try {
        var a = 0; // counter for index
        for (var b = this.getStart(); b !== null; b = b.getNext()) // iterate over all cells
            b.$f = a++; // assign index to cell and then increment
    }
    catch (e) { throw e }
    this.$f5 = !1; // random code
};

This is at least a documented behavior of the current version of Google's V8 (see the Google I/O presentation on V8), but I don't know for Safari.

If you want to disable it for the whole script, one solution would be to compile your JS to wrap every function's content inside a try-catch with a tool such as burrito.

Good job on making this reproducible!

回答2:

IMO, the correct solution is to report the bug to Apple, then workaround it in your code (surely using a separate a = a + 1; statement will work, unless the JIT is even worse than you thought!). It does indeed suck, though. Here's a list of common things you can also try throwing in to the function to make it de-optimise and not use JIT:

Exceptions
'with' statement
using arguments object, e.g. arguments.callee
eval()

The problem with those is if the Javascript engine is optimised to JIT them before they fix that bug, in which case you're back to crashing. So, report and workaround!

回答3:

Actually, the FOR loop bug is still present in Safari on iOS 7.0.4 in iPhone 4 and iPad 2. The loop failing can be significantly simpler than the illustration above, and it rakes several passes through the code to hit. Changing to a WHILE loop allows proper execution.

Failing code:

function zf(num,digs) 
{ 
var out = ""; 
var n = Math.abs(num); 
for (digs;  digs>0||n>0; digs--)
{ 
    out = n%10 + out; 
    n = Math.floor(n/10); 
}  
return num<0?"-"+out:out; 
}

Successful code:

function zf(num,digs) 
{ 
var out = ""; 
var n = Math.abs(num); 
do 
{ 
    out = n%10 + out; 
    n = Math.floor(n/10); 
} 
while (--digs>0||n>0) 
return num<0?"-"+out:out; 
}