fail - Dave's Blog

Its rare to find devs anticipating Unicode control characters showing up in user input. And the most fun when unanticipated is the Right-To-Left Override character U+202E. Unicode characters have an implicit direction so that for example by default Hebrew characters are rendered from right to left, and English characters are rendered left to right. The override characters force an explicit direction for all the text that follows.

I chose my Twitter display name to include the HTML encoding of the Right-To-Left Override character #x202E; as a sort of joke or shout out to my favorite Unicode control character. I did not anticipate that some Twitter clients in some of their UI would fail to encode it correctly. There's no way I can remove that from my display name now.

Try it on Amazon.

JSBrowser is a basic browser built as a Win10 JavaScript UWP app around the WebView HTML element. Its fun and relatively simple to implement tiny browser features in JavaScript and in this post I'm implementing crash resistance.

The normal DOM mechanisms for creating an HTML WebView create an in-process WebView, in which the WebView runs on a unique UI thread. But we can use the MSWebView constructor instead to create an out-of-process WebView in which the WebView runs in its own distinct WebView process. Unlike an in-process WebView, Web content running in an out-of-process WebView can only crash the WebView process and not the app process.

        this.replaceWebView = () => {
            let webview = document.querySelector("#WebView");
            // Cannot access webview.src - anything that would need to communicate with the webview process may fail
            let oldSrc = browser.currentUrl;
            const webviewParent = webview.parentElement;
            webviewParent.removeChild(webview);
            webview = new MSWebView();
            Object.assign(this, {
                "webview": webview
            });
            webview.setAttribute("id", "WebView");

            // During startup our currentUrl field is blank. If the WebView has crashed 
            // and we were on a URI then we may obtain it from this property.
            if (browser.currentUrl && browser.currentUrl != "") {
                this.trigger("newWebview");
                this.navigateTo(browser.currentUrl);
            }
            webviewParent.appendChild(webview);

I run replaceWebView during startup to replace the in-process WebView created via HTML markup with an out-of-process WebView. I could be doing more to dynamically copy styles, attributes, etc but I know what I need to set on the WebView and just do that.

When a WebView process crashes the corresponding WebView object is no longer useful and a new WebView element must be created. In fact if the old WebView object is used it may throw and will no longer have valid state. Accordingly when the WebView crashes I run replaceWebView again. Additionally, I need to store the last URI we've navigated to (browser.currentUrl in the above) since the crashed WebView object won't know what URI it is on after it crashes.

            webview.addEventListener("MSWebViewProcessExited", () => { 
                if (browser.currentUrl === browser.lastCrashUrl) {                   ++browser.lastCrashUrlCrashCount; 
                } 
                else { 
                    browser.lastCrashUrl = browser.currentUrl; 
                    browser.lastCrashUrlCrashCount = 1; 
                } 
                // If we crash again and again on the same URI, maybe stop trying to load that URI. 
                if (browser.lastCrashUrlCrashCount >= 3) { 
                    browser.lastCrashUrl = ""; 
                    browser.lastCrashUrlCrashCount = 0; 
                    browser.currentUrl = browser.startPage; 
                } 
                this.replaceWebView(); 
            }); 

I also keep track of the last URI that we recovered and how many times we've recovered that same URI. If the same URI crashes more than 3 times in a row then I assume that it will keep happening and I navigate to the start URI instead.

Win10 Changes

In Win8.1 JavaScript UWP apps we supported multiple windows using MSApp DOM APIs. In Win10 we use window.open and window and a new MSApp API getViewId and the previous MSApp APIs are gone:

	Win10	Win8.1
Create new window	window.open	MSApp.createNewView
New window object	window	MSAppView
viewId	MSApp.getViewId(window)	MSAppView.viewId

WinRT viewId

We use window.open and window for creating new windows, but then to interact with WinRT APIs we add the MSApp.getViewId API. It takes a window object as a parameter and returns a viewId number that can be used with the various Windows.UI.ViewManagement.ApplicationViewSwitcher APIs.

Delaying Visibility

Views in WinRT normally start hidden and the end developer uses something like TryShowAsStandaloneAsync to display the view once it is fully prepared. In the web world, window.open shows a window immediately and the end user can watch as content is loaded and rendered. To have your new windows act like views in WinRT and not display immediately we have added a window.open option. For example
let newWindow = window.open("https://example.com", null, "msHideView=yes");

Primary Window Differences

The primary window that is initially opened by the OS acts differently than the secondary windows that it opens:

	Primary	Secondary
window.open	Allowed	Disallowed
window.close	Close app	Close window
Navigation restrictions	ACUR only	No restrictions

The restriction on secondary windows such that they cannot open secondary windows could change in the future depending on feedback.

Same Origin Communication Restrictions

Lastly, there is a very difficult technical issue preventing us from properly supporting synchronous, same-origin, cross-window, script calls. That is, when you open a window that's same origin, script in one window is allowed to directly call functions in the other window and some of these calls will fail. postMessage calls work just fine and is the recommended way to do things if that's possible for you. Otherwise we continue to work on improving this.

The x-ms-webview HTML element has the void addWebAllowedObject(string name, any value) method and the webview XAML element has the void AddWebAllowedObject(String name, Object value) method. The object parameter is projected into the webview’s top-level HTML document’s script engine as a new property on the global object with property name set to the name parameter. It is not injected into the current document but rather it is projected during initialization of the next top-level HTML document to which the webview navigates.

Lifetime

If AddWebAllowedObject is called during a NavigationStarting event handler the object will be injected into the document resulting from the navigation corresponding to that event.

If AddWebAllowedObject is called outside of the NavigationStarting event handler it will apply to the navigation corresponding to the next explicit navigate method called on the webview or the navigation corresponding to the next NavigationStarting event handler that fires, whichever comes first.

To avoid this potential race, you should use AddWebAllowedObject in one of two ways: 1. During a NavigationStarting event handler, 2. Before calling a Navigate method and without returning to the main loop.

If called both before calling a navigate method and in the NavigationStarting event handler then the result is the aggregate of all those calls.

If called multiple times for the same document with the same name the last call wins and the previous are silently ignored.

If AddWebAllowedObject is called for a navigation and that navigation fails or redirects to a different URI, the AddWebAllowedObject call is silently ignored.

After successfully adding an object to a document, the object will no longer be projected once a navigation to a new document occurs.

WinRT access

If AddWebAllowedObject is called for a document with All WinRT access then projection will succeed and the object will be added.

If AddWebAllowedObject is called for a document which has a URI which has no declared WinRT access via ApplicationContentUriRules then Allow for web only WinRT access is given to that document.

If the document has Allow for web only WinRT access then projection will succeed only if the object’s runtimeclass has the Windows.Foundation.Metadata.AllowForWeb metadata attribute.

Object requirements

The object must implement the IAgileObject interface. Because the XAML and HTML webview elements run on ASTA view threads and the webview’s content’s JavaScript thread runs on another ASTA thread a developer should not create their non-agile runtimeclass on the view thread. To encourage end developers to do this correctly we require the object implements IAgileObject.

Property name

The name parameter must be a valid JavaScript property name, otherwise the call will fail silently. If the name is already a property name on the global object, that property is overwritten if the property is configurable. Non-configurable properties on the global object are not overwritten and the AddWebAllowedObject call fails silently. On success, the projected property is writable, configurable, and enumerable.

Errors

Some errors as described above fail silently. Other issues, such as lack of IAgileObject or lack of the AllowForWeb attribute result in an error in the JavaScript developer console.

JavaScript Microsoft Store apps have some details related to activation that are specific to JavaScript Store apps and that are poorly documented which I’ll describe here.

StartPage syntax

The StartPage attributes in the AppxManifest.xml (Package/Applications/Application/@StartPage, Package/Applications/Extensions/Extension/@StartPage) define the HTML page entry point for that kind of activation. That is, Application/@StartPage defines the entry point for tile activation, Extension[@Category="windows.protocol"]/@StartPage defines the entry point for URI handling activation, etc. There are two kinds of supported values in StartPage attributes: relative Windows file paths and absolute URIs. If the attribute doesn’t parse as an absolute URI then it is instead interpreted as relative Windows file path.

This implies a few things that I’ll declare explicitly here. Windows file paths, unlike URIs, don’t have a query or fragment, so if you are using a relative Windows file path for your StartPage attribute you cannot include anything like ‘?param=value’ at the end. Absolute URIs use percent-encoding for reserved characters like ‘%’ and ‘#’. If you have a ‘#’ in your HTML filename then you need to percent-encode that ‘#’ for a URI and not for a relative Windows file path.

If you specify a relative Windows file path, it is turned into an ms-appx URI by changing all backslashes to forward slashes, percent-encoding reserved characters, and combining the result with a base URI of ms-appx:///. Accordingly the relative Windows file paths are relative to the root of your package. If you are using a relative Windows file path as your StartPage and need to switch to using a URI so you can include a query or fragment, you can follow the same steps above.

StartPage validity

The validity of the StartPage is not determined before activation. If the StartPage is a relative Windows file path for a file that doesn’t exist, or an absolute URI that is not in the Application Content URI Rules, or something that doesn’t parse as a Windows file path or URI, or otherwise an absolute URI that fails to resolve (404, bad hostname, etc etc) then the JavaScript app will navigate to the app’s navigation error page (perhaps more on that in a future blog post). Just to call it out explicitly because I have personally accidentally done this: StartPage URIs are not automatically included in the Application Content URI Rules and if you forget to include your StartPage in your ACUR you will always fail to navigate to that StartPage.

StartPage navigation

When your app is activated for a particular activation kind, the StartPage value from the entry in your app’s manifest that corresponds to that activation kind is used as the navigation target. If the app is not already running, the app is activated, navigated to that StartPage value and then the Windows.UI.WebUI.WebUIApplication activated event is fired (more details on the order of various events in a moment). If, however, your app is already running and an activation occurs, we navigate or don’t navigate to the corresponding StartPage depending on the current page of the app. Take the app’s current top level document’s URI and if after removing the fragment it already matches the StartPage value then we won’t navigate and will jump straight to firing the WebUIApplication activated event.

Since navigating the top-level document means destroying the current JavaScript engine instance and losing all your state, this behavior might be a problem for you. If so, you can use the MSApp.pageHandlesAllApplicationActivations(true) API to always skip navigating to the StartPage and instead always jump straight to firing the WebUIApplication activated event. This does require of course that all of your pages all handle all activation kinds about which any part of your app cares.

The other day I had to debug a JavaScript UWA that was failing when trying to use an undefined property. In a previous OS build this code would run and the property was defined. I wanted something similar to windbg/cdb's ba command that lets me set a breakpoint on read or writes to a memory location so I could see what was creating the object in the previous OS build and what that code was doing now in the current OS build. I couldn't find such a breakpoint mechanism in Visual Studio or F12 so I wrote a little script to approximate JavaScript data breakpoints.

The script creates a stub object with a getter and setter. It actually performs the get or set but also calls debugger; to break in the debugger. In order to handle my case of needing to break when window.object1.object2 was created or accessed, I further had it recursively set up such stub objects for the matching property names.

Its not perfect because it is an enumerable property and shows up in hasOwnProperty and likely other places. But for your average code that checks for the existence of a property via if (object.property) it works well.

Some time back while I was working on getting the Javascript Windows Store app platform running on Windows Phone (now available on the last Windows Phone release!) I had an interesting bug that in retrospect is amusing.

I had just finished a work item to get accessibility working for JS WinPhone apps when I got a new bug: With some set of JS apps, accessibility appeared to be totally broken. At that time in development the only mechanism we had to test accessibility was a test tool that runs on the PC, connects to the phone, and dumps out the accessibility tree of whatever app is running on the phone. In this bug, the tool would spin for a while and then timeout with an error and no accessibility information.

My first thought was this was an issue in my new accessibility code. However, debugging with breakpoints on my code I could see none of my code was run nor the code that should call it. The code that called that code was a more generic messaging system that hit my breakpoints constantly.

Rather than trying to work backward from the failure point, I decided to try and narrow down the repro and work forwards from there. One thing all the apps with the bug had in common was their usage of WinJS, but not all WinJS apps demonstrated the issue. Using a binary search approach on one such app I removed unrelated app code until all that was left was the app's usage of the WinJS AppBar and the bug still occurred. I replaced the WinJS AppBar usage with direct usage of the underlying AppBar WinRT APIs and continued.

Only some calls to the AppBar WinRT object produced the issue:

        var appBar = Windows.UI.WebUI.Core.WebUICommandBar.getForCurrentView(); 
        // appBar.opacity = 1;
        // appBar.closeDisplayMode = Windows.UI.WebUI.Core.WebUICommandBarClosedDisplayMode.default;
        appBar.backgroundColor = Windows.UI.Colors.white; // Bug! 

Just setting the background color appeared to cause the issue and I didn't even have to display the AppBar. Through additional trial and error I was blown away to discover that some colors I would set caused the issue and other colors did not. Black wouldn't cause the issue but transparent black would. So would aqua but not white.

I eventually realized that predefined WinRT color values like Windows.UI.Colors.aqua would cause the issue while JS literal based colors didn't cause the issue (Windows.UI.Color is a WinRT struct which projects in JS as a JS literal object with the struct members as JS object properties so its easy to write something like {r: 0, g: 0, b: 0, a: 0} to make a color) and I had been mixing both in my tests without realizing there would be a difference. I debugged into the backgroundColor property setter that consumed the WinRT color struct to see what was different between Windows.UI.Colors.black and {a: 1, r: 0, g: 0, b: 0} and found the two structs to be byte wise exactly the same.

On a hunch I tried my test app with only a reference to the color and otherwise no interaction with the AppBar and not doing anything with the actual reference to the color: Windows.UI.Colors.black;. This too caused the issue. I knew that the implementation for these WinRT const values live in a DLL and guessed that something in the code to create these predefined colors was causing the issue. I debugged in and no luck. Now I also have experienced crusty code that would do exciting things in its DllMain, the function that's called when a DLL is loaded into the process so I tried modifying my C++ code to simply LoadLibrary the DLL containing the WinRT color definition, windows.ui.xaml.dll and found the bug still occurred! A short lived moment of relief as the world seemed to make sense again.

Debugging into DllMain nothing interesting happened. There were interesting calls in there to be sure, but all of them behind conditions that were false. I was again stumped. On another hunch I tried renaming the DLL and only LoadLibrary'ing it and the bug went away. I took a different DLL renamed it windows.ui.xaml.dll and tried LoadLibrary'ing that and the bug came back. Just the name of the DLL was causing the issue.

I searched for the DLL name in our source code index and found hits in the accessibility tool. Grinning I opened the source to find that the accessibility tool's phone side service was trying to determine if a process belonged to a XAML app or not because XAML apps had a different accessibility contract. It did this by checking to see if windows.ui.xaml.dll was loaded in the target process.

At this point I got to fix my main issue and open several new bugs for the variety of problems I had just run into. This is a how to on writing software that is difficult to debug.

Level 8 of the Stripe CTF is a password server that returns success: true if and only if the password provided matches the password stored directly via a RESTful API and optionally indirectly via a callback URI. The solution is side channel attack like a timing attack but with ports instead of time.

(I found this in my drafts folder and had intended to post a while ago.)

Code

    def nextServerCallback(self, data):
        parsed_data = json.loads(data)
        # Chunk was wrong!
        if not parsed_data['success']:
            # Defend against timing attacks
            remaining_time = self.expectedRemainingTime()
            self.log_info('Going to wait %s seconds before responding' %
                          remaining_time)
            reactor.callLater(remaining_time, self.sendResult, False)
            return

        self.checkNext()

Issue

The password server breaks the target password into four pieces and stores each on a different server. When a password request is sent to the main server it makes requests to the sub-servers for each part of the password request. It does this in series and if any part fails, then it stops midway through. Password requests may also be made with corresponding URI callbacks and after the server decides on the password makes an HTTP request on the provided URI callbacks saying if the password was success: true or false.
A timing attack looks at how long it took for a password to be rejected and longer times could mean a longer prefix of the password was correct allowing for a directed brute force attack. Timing attacks are prevented in this case by code on the password server that attempts to wait the same amount of time, even if the first sub-server responds with false. However, the server uses sequential outgoing port numbers shared between the requests to the sub-servers and the callback URIs. Accordingly, we can examine the port numbers on our callback URIs to direct a brute force attack.
If the password provided is totally incorrect then the password server will contact one sub-server and then your callback URI. So if you see the remote server's port number go up by two when requesting your callback URI, you know the password is totally incorrect. If by three then you know the first fourth of the password is correct and the rest is incorrect. If by four then two fourths of the password is correct. If by five then four sub-servers were contacted so you need to rely on the actual content of the callback URI request of 'success: true' or 'false' since you can't tell from the port change if the password was totally correct or not.
The trick in the real world is false positives. The port numbers are sequential over the system, so if the password server is the only thing making outgoing requests then its port numbers will also be sequential, however other things on the system can interrupt this. This means that the password server could contact three sub-servers and normally you'd see the port number increase by four, but really it could increase by four or more because of other things running on the system. To counteract this I ran in cycles: brute forcing the first fourth of the password and removing any entry that gets a two port increase and keeping all others. Eventually I could remove all but the correct first fourth of the password. And so on for the next parts of the password.
I wrote my app to brute force this in Python. This was my first time writing Python code so it is not pretty.

One of the more limiting issues of writing client side script in the browser is the same origin limitations of XMLHttpRequest. The latest version of all browsers support a subset of CORS to allow servers to opt-in particular resources for cross-domain access. Since IE8 there's XDomainRequest and in all other browsers (including IE10) there's XHR L2's cross-origin request features. But the vast majority of resources out on the web do not opt-in using CORS headers and so client side only web apps like a podcast player or a feed reader aren't doable.

One hack-y way around this I've found is to use YQL as a CORS proxy. YQL applies the CORS header to all its responses and among its features it allows a caller to request an arbitrary XML, HTML, or JSON resource. So my network helper script first attempts to access a URI directly using XDomainRequest if that exists and XMLHttpRequest otherwise. If that fails it then tries to use XDR or XHR to access the URI via YQL. I wrap my URIs in the following manner, where type is either "html", "xml", or "json":

        yqlRequest = function(uri, method, type, onComplete, onError) {
            var yqlUri = "http://query.yahooapis.com/v1/public/yql?q=" + 
                encodeURIComponent("SELECT * FROM " + type + ' where url="' + encodeURIComponent(uri) + '"');

            if (type == "html") {
                yqlUri += encodeURIComponent(" and xpath='/*'");
            }
            else if (type == "json") {
                yqlUri += "&callback=&format=json";
            }
            ...

This also means I can get JSON data itself without having to go through JSONP.

As a professional URI aficionado I deal with various levels of ignorance on URI percent-encoding (aka URI encoding, or URL escaping).

Worse than the lame blog comments hating on percent-encoding is the shipping code which can do actual damage. In one very large project I won't name, I've fixed code that decodes all percent-encoded octets in a URI in order to get rid of pesky percents before calling ShellExecute. An unnamed developer with similar intent but clearly much craftier did the same thing in a loop until the string's length stopped changing. As it turns out percent-encoding serves a purpose and can't just be removed arbitrarily.

Percent-encoding exists so that one can represent data in a URI that would otherwise not be allowed or would be interpretted as a delimiter instead of data. For example, the space character (U+0020) is not allowed in a URI and so must be percent-encoded in order to appear in a URI:

1. http://example.com/the%20path/
2. http://example.com/the path/

In the above the first is a valid URI while the second is not valid since a space appears directly in the URI. Depending on the context and the code through which the wannabe URI is run one may get unexpected failure.

For an additional example, the question mark delimits the path from the query. If one wanted the question mark to appear as part of the path rather than delimit the path from the query, it must be percent-encoded:

1. http://example.com/foo%3Fbar
2. http://example.com/foo?bar

In the second, the question mark appears plainly and so delimits the path "/foo" from the query "bar". And in the first, the querstion mark is percent-encoded and so the path is "/foo%3Fbar".