Abstract

This document defines directives for the Content Security Policy mechanism to declare a set of input protections for a web resource's user interface, defines a non-normative set of heuristics for Web user agents to implement these input protections, and a reporting mechanism for when they are triggered.

Status of This Document

This section describes the status of this document at the time of its publication. Other documents may supersede this document. A list of current W3C publications and the latest revision of this technical report can be found in the W3C technical reports index at http://www.w3.org/TR/.

This is a Working Draft of the User Interface Security Directives for Content Security Policy. [CSP]

Portions of the technology described in this document were originally developed as part of X-Frame-Options [RFC7034], the ClearClick module of the Mozilla Firefox add-on NoScript, [CLEARCLICK] and in the InContext system implemented experimentally in Internet Explorer [INCONTEXT].

In addition to the documents in the W3C Web Application Security working group, the work on this document is also informed by the work of the IETF websec working group, particularly that working group's requirements document: draft-hodges-websec-framework-reqs.

This document was published by the Web Application Security Working Group as a Last Call Working Draft. This document is intended to become a W3C Recommendation. If you wish to make comments regarding this document, please send them to public-webappsec@w3.org (subscribe, archives). The Last Call period ends 18 June 2014. All comments are welcome.

Publication as a Last Call Working Draft does not imply endorsement by the W3C Membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than work in progress.

This is a Last Call Working Draft and thus the Working Group has determined that this document has satisfied the relevant technical requirements and is sufficiently stable to advance through the Technical Recommendation process.

This document was produced by a group operating under the 5 February 2004 W3C Patent Policy. W3C maintains a public list of any patent disclosures made in connection with the deliverables of the group; that page also includes instructions for disclosing a patent. An individual who has actual knowledge of a patent which the individual believes contains Essential Claim(s) must disclose the information in accordance with section 6 of the W3C Patent Policy.

Table of Contents

1. Introduction

This section is non-normative.

This document defines User Interface Security directives for Content Security Policy, a mechanism web applications can use to mitigate some of the risks of User Interface (UI) Redressing [UIREDRESS] (AKA "Clickjacking") vulnerabilities that can lead to fraudulent actions not intended by the user.

Content Security Policy (CSP) is a declarative policy that lets the authors (or server administrators) of a web application restrict the behavior of a document, e.g. the origins where it can load its resources from or the ways it can execute scripts. This document defines directives to restrict the presentation or the interactivity of a resource when its interaction with the user may be happening in an ambiguous or deceitful context due to the spatial and/or temporal contiguity with other content displayed by the user agent.

A user agent may implement the core directives of CSP independently from the directives in this specification, but this specification requires the policy conveyance and reporting mechanisms described in CSP. The interpretation of terms imported into this document from CSP may vary depending on the version implemented by the user agent. For example, a source-expression in Content Security Policy 1.0 is at the granularity of an origin [ORIGIN] but may be more granular in future versions of the core Content Security Policy.

Application authors SHOULD transmit the directives in this specification as part of a single, complete Content Security Policy, as indicated by that specification.

In some UI Redressing attacks (also known as Clickjacking), a malicious web application presents a user interface of another web application in a manipulated context to the user, e.g. by partially obscuring the genuine user interface with opaque layers on top, hence tricking the user to click on a button out of context.

Existing anti-clickjacking measures including frame-busting [FRAMEBUSTING] codes and X-Frame-Options cannot be used to protect resources where the set of origins that should be allowed and disallowed is unknown, where attacks might come from origins intended to be allowed by a use scenario, or defend against timing-based attacks involving multiple windows instead of multiple frames. Frame-busting scripts also rely on browser behavior that has not been engineered to provide a security guarantee. As a consequence, such scripts may be unreliable if loaded inside a sandbox or otherwise disabled.

The User Interface Security directives encompass the policies defined in X-Frame-Options and also provide a new mechanism to allow web applications to enable heuristic input protections for its user interfaces on user agents.

To mitigate UI redressing, for example, a web application can request that a user interface element should be fully visible for a minimum period of time before a user input can be delivered.

The User Interface Security directive can often be applied to existing applications with few or no changes, but the heuristic hints supplied by the policy may require considerable experimental fine-tuning to achieve an acceptable error rate.

This specification supercedes X-Frame-Options. Resources may supply an X-Frame-Options header in addition to a Content-Security-Policy header to indicate policy to user agents that do not implement the directives in this specification. A user agent that understands the directives in this document SHOULD ignore the X-Frame-Options header, when present, if User Interface Security directives are also present in a Content-Security-Policy header. This is to allow resources to only be embedded if the mechanisms described in this specification are enforced, and more restrictive X-Frame-Options policies applied otherwise.

2. Conformance

As well as sections marked as non-normative, all authoring guidelines, diagrams, examples, and notes in this specification are non-normative. Everything else in this specification is normative.

The key words MUST, MUST NOT, REQUIRED, SHOULD, SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL in this specification are to be interpreted as described in [RFC2119].

Requirements phrased in the imperative as part of algorithms (such as "strip any leading space characters" or "return false and abort these steps") are to be interpreted with the meaning of the key word ("MUST", "SHOULD", "MAY", etc) used in introducing the algorithm.

A conformant user agent is one that implements all the requirements listed in this specification that are applicable to user-agents. Treatment of the input-protection, input-protection-clip and input-protection-selectors directives are at the discretion of the user agent.

A conformant server is one that implements all the requirements listed in this specification that are applicable to servers.

2.1 Terminology

This section defines several terms used throughout the document.

The term security policy, or simply policy, for the purposes of this specification refers to either:

  1. a set of security preferences for restricting the behavior of content within a given resource, or
  2. a fragment of text that codifies these preferences.

The security policies defined by this document are applied by a user agent on a per-resource representation basis. Specifically, when a user agent receives a policy along with the representation of a given resource, that policy applies to that resource representation only. This document often refers to that resource representation as the protected resource.

A server transmits its security policy for a particular resource as a collection of directives, such as default-src 'self', each of which controls a specific set of privileges for a document rendered by the user agent. More details are provided in the directives section.

A directive consists of a directive name, which indicates the privileges controlled by the directive, and a directive value, which specifies the restrictions the policy imposes on those privileges.

An ancestor is any resource between the protected resource and the top of the window frame tree; for example, if A embeds B which embeds C, both A and B are ancestors of C. If A embeds both B and C, B is not an ancestor of C, but A still is.

The term origin is defined in the Origin specification. [ORIGIN]

The term URI is defined in the URI specification. [URI]

The <iframe>, <object>, <embed>, and <frame> elements are defined in the HTML5 standard. [HTML5].

The <applet> element is defined in the HTML 4.01 standard. [HTML401].

The Augmented Backus-Naur Form (ABNF) notation used in this document is specified in RFC 5234. [ABNF]

The following core rules are included by reference, as defined in [ABNF Appendix B.1]: ALPHA (letters), DIGIT (decimal 0-9), WSP (white space) and VCHAR (printing characters).

The OWS rule is used where zero or more linear whitespace octets might appear. OWS SHOULD either not be produced or be produced as a single SP. Multiple OWS octets that occur within field-content SHOULD either be replaced with a single SP or transformed to all SP octets (each octet other than SP replaced with SP) before interpreting the field value or forwarding the message downstream.

OWS            = *( SP / HTAB / obs-fold )
               ; "optional" whitespace
obs-fold       = CRLF ( SP / HTAB )
               ; obsolete line folding

A selector string is a list of one or more complex selectors(see [SELECTORS4], section 3.1) that may be surrounded by whitespace and matches the dom_selectors_group production.

dom_selectors_group
  : S* [ selectors_group ] S*
  ;

A embedding source list follows the ABNF and parsing rules defined for source-list (see [CSP] section 3.22) with the following new productions:

embedding-keyword-source = "'self'" / "'deny'"
embedding-source-expression = host-source / embedding-keyword-source
embedding-source-list = *WSP [ embedding-source-expression *( 1*WSP embedding-source-expression ) *WSP ]

3. Directives

This section describes the content security policy directives introduced in this specification.

4. input-protection

The input-protection directive, if present or implied, instructs the user agent to apply the heuristic UI redressing protections described in the Input Protection Heuristic section to user input events, such as click, keypress, touch, and drag, before they are delivered to the resource.

The screenshot comparison heuristic, in particular, uses the body-bounding rectangle of the document triggering the event as its default reference area, or the rectangle defined by the input-protection-clip and by the input-protection-selectors directives if any of those is explicitly set.

If the input-protection-clip directive is set as part of a Content-Security-Policy, triggering of the heuristic should cancel delivery of the UI event to the target and cause a violation report to be sent. If set as part of a Content-Security-Policy-Report-Only, triggering of the heuristic should result in the event being delivered with the unsafe attribute on the UIEvent set to true and cause a violation report to be sent.

The optional directive value allows resource authors to provide options for heuristic tuning in the form of space-separated option-name=option-value pairs.

directive-name    = "input-protection"
directive-value   = ["display-time=" num-val] ["tolerance=" num-val]

If the policy does not contain a value for this directive or any of the hint name=value pairs are absent, the user agent SHOULD apply default values for hints as described in the following.

display-time
is a numeric value from 0 to 10000 that specifies how long, in milliseconds, the screen area containing the protected user interface must have been displayed continuously unchanged when the event is processed. If not specified, it defaults to 800. If a value out of the range stated above is specified, it defaults to the nearest value between the lower and the higher bounds.
tolerance
is a numeric value from 0 to 99 that defines the difference threshold at which the screenshot comparison procedure of the input protection heuristic triggers a violation. A value of 0 indicates that no difference between the two images is permitted. A value of 99 provides little to no practical protection. If not specified, it defaults to 0.

5. input-protection-clip

The input-protection-clip directive defines a rectangular screen area whose intersection with the bounding rectangle of the whole document's body should be used as the reference area in the screenshot comparison check explained in the Input Protection Heuristic section.

If the input-protection-clip directive is not explicitly set in a policy which includes the input-protection directive and no input-protection-selectors directive is set either, the bounding rectangle of the whole document's body should be used for screenshot comparisons.

If explicitly set as part of a policy where no input-protection directive is explicitly set, the input-protection-clip directive implies the input-protection directive as if it was set in the same policy with its default value.

directive-name  = "input-protection-clip"
directive-value = ["before=" num-val] ["above=" num-val] ["after=" num-val] ["below=" num-val]

The optional directive value can include up to four non-negative numeric labeled offsets, expressed in CSS pixels and relative to the screen coordinates of the UI event being processed (event.screenX and event.screenY for mouse, touch or pointer events) or, if not applicable (e.g. for keyboard events), to the geometrical center of the event target in screen coordinates. These offsets define a rectangle with

x = eX - left, y = eY - top, width = left + right, height = top + bottom
where eX and eY are the event's explicit (when possible) or inferred (the target's center) screen ordinates. The left, top, right and bottom values are mapped to the offsets labeled as before, above, after and below respectively, unless the bi-directional text properties of the event target suggest otherwise: for instance, if the target's direction is RTL, before translates to right and after translates to left.

The default value for this directive is before=250 above=250 after=50 below=50. If a partial value is provided (i.e. any offset has been omitted) the default values should be implied for the missing offsets.

The intersection of the computed rectangle with the bounding rectangle of the document's body should be used as the reference area for the screenshot comparison check explained in the Input Protection Heuristic section, unless the UI event's target or one of its DOM ancestors match a input-protection-selector directive set in the same policy.

If the input-protection-clip directive is not set or provides an invalid value, the whole bounding rectangle of the document's body must be used as the reference area for the screenshot comparison, unless an input-protection-selectors directive is set in the same policy.

6. input-protection-selectors

The input-protection-selectors directive overrides the implicit or explicit input-protection-clip value when the processed UI event target or one of its DOM ancestors match the dom_selectors_group selector string provided as the mandatory directive's value: in this case, the reference area used for screenshot comparison is the bounding box of the event target itself, if it matches the selectors, or the bounding box of its nearest matching DOM ancestor, if any, augmented by the margins given by the leading optional labeled offsets, if any. UI events whose target and ancestors don't match any of the specified selectors should be ignored (not blocked) unless an input-protection-clip directive is explicitly included in the policy: if this is the case, the UI event must be checked and the screenshot reference area should be computed using the input-protection-clip directive.

If set as part of a policy where no input-protection directive is explicitly set, the input-protection-selectors directive implies the input-protection directive as if it was set in the same policy with its default value.

directive-name  = "input-protection-selectors"
directive-value = ["before=" num-value] ["after=" num-value] ["above=" num-value] ["below=" num-value] dom_selectors_group

Any of the four non-negative numeric labeled offsets, which represent margins expressed in CSS pixels, may be omitted, taking 0 (zero) as their default values.

The reference screenshot area is computed as the rectangle having

x = match.x - left, y = match.y - top, width = left + match.width + right, height = top + match.height + bottom
where match is the bounding rectangle around the UI event target, if it matches dom_selectors_group, or around its nearest matching ancestor. The left, top, right and bottom values are mapped to the offsets labeled as before, above, after and below respectively, unless the bi-directional text properties of the event target suggest otherwise: for instance, if the target's direction is RTL, before translates to right and after translates to left (similarly to the input-protection-clip directive).

7. report-uri

The report-uri directive specifies a URI to which the user agent sends reports about policy violation.

The syntax for the name and value of this directive and the algorithm to prepare a report are described by Content Security Policy. [CSP]

The core Content Security Policy specification provides directives to restrict from where external content may be loaded. As such, violation reports include a blocked-uri key/value pair that specifies the attempted resource load that was blocked by the policy.

As this is not applicable to the directives in this document, the following additional steps MUST be added to the algorithm defined in Content Security Policy to prepare a violation report:

In step 1, when preparing the JSON object violation-object, add the following keys and values to the csp-report: [CSP]

If the violation is of the input-protection directive, add the following keys and values. If a value is not set or applicable for the violation (e.g. pointer-height, if the violating event type is not a Pointer Event) the key SHOULD be omitted.

blocked-event-type
The type attribute of the UIEvent that was blocked by policy.
touch-event
A boolean indicating whether the event blocked by policy was a Touch Event [TOUCH-EVENTS].
pointer-type
The pointerType value of a Pointer Event [POINTER-EVENTS].
pointer-height
The height value of a Pointer Event.
pointer-width
The width value of a Pointer Event.
device-height
The device-height property as defined in [CSS3-MEDIAQUERIES].
device-width
The device-width property as defined in [CSS3-MEDIAQUERIES].
blocked-event-client-x
The clientX attribute of the UIEvent [DOM-LEVEL-2-EVENTS] that was blocked by policy, if set.
blocked-event-client-y
The clientY attribute of the UIEvent [DOM-LEVEL-2-EVENTS] that was blocked by policy, if set.

If the target of an UIEvent which triggers an input-protection violation has an explictly-set id attribute:

blocked-target-id
The id attribute of the DOM Element that a violating UIEvent targeted.

Otherwise, if the target element does not have an explicit id attribute:

blocked-target-xpath
An XPath [XPATH] expression that returns the target Element of the UIEvent that was blocked by policy.

7.1 Producing blocked-target-xpath

This section is non-normative.

User agent implementers may provide any unambiguous XPath in the report. The following example code using the ECMAScript language bindings for DOM Level 2 Core [DOM-LEVEL-2-CORE] produces an unambiguous XPath to the target DOM element "e":
Example 1: Sample implementation of XPath generation for reporting
function getXPathFor(e) {
 
    var xpath = '';
    
    while(e.nodeType == e.ELEMENT_NODE) {
      
      var child = e;
      var siblingIndex = 0;
      while( (child = child.previousSibling) != null ) {
        if(child.tagName == e.tagName) {
          siblingIndex++;  
        }
      }
        
      xpath = e.tagName + 
              '[' + siblingIndex + ']' + 
              (xpath == '' ? '' : '/') +
              xpath;
        
      e = e.parentNode;
   }
   xpath = '/' + xpath; 
   return(xpath);
}
Documents may be dynamically constructed and change structure in response to user interaction or other events, so an unambiguous XPath expression in the context of the current state of the DOM may not be unambiguous to the content author. To avoid this confusion, resource authors SHOULD include an id attribute for all elements of interest and user agent implementers MAY include any additional information in the XPath they feel may help disambiguate the blocked target, including class names and id attributes of ancestors.

8. DOM interface

This specification introduces a new attribute for the UIEvent interface introduced in DOM Level 2. [DOM-LEVEL-2-EVENTS]

partial interface UIEvent {
    [Unforgeable]
    readonly    attribute bool unsafe;
};

Attributes

unsafe of type bool, readonly
This is a non-configurable boolean property of input event objects. Set to "true" if a violation of an input-protection directive violation occurred for the event.

The unsafe attribute allows web applications to monitor and immediately respond to suspect violations in the report-only mode. Applications may also use this interface for capability detection. For example, a web application may monitor user inputs on a payment button element like this:

Example 2: Example code responing to unsafe attribute
document.getElementById('payment-button').addEventListener("click", function(eventObj) {
  if ("unsafe" in eventObj) {
    if (eventObj.unsafe == true) {
      return reportUnsafeOrShowDialog();
    }
  }
  makePayment();
};

9. Script Interfaces

If associated with a Content Security Policy 1.1 [CSP11] or later implementation, the User Interface Security Directives include the following script interfaces which extend the experimental functinality defined therein: https://dvcs.w3.org/hg/content-security-policy/raw-file/tip/csp-specification.dev.html#script-interfaces--experimental

9.1 SecurityPolicyViolationEvent Events

[Constructor(DOMString type, optional SecurityPolicyViolationEventInit eventInitDict)]
partial interface SecurityPolicyViolationEvent : Event {
    readonly    attribute DOMString blockedEventType;
    readonly    attribute bool      touchEvent;
    readonly    attribute DOMString pointerType;
    readonly    attribute long      pointerHeight;
    readonly    attribute long      pointerWidth;
    readonly    attribute long      deviceHeight;
    readonly    attribute long      deviceWidth;
    readonly    attribute long      blockedEventClientX;
    readonly    attribute long      blockedEventClientY;
    readonly    attribute DOMString blockedTargetID;
    readonly    attribute DOMString blockedTargetXPath;
};

9.1.1 Attributes

blockedEventClientX of type long, readonly
Refer to the blocked-event-client-x property of violation reports for a description of this property.
blockedEventClientY of type long, readonly
Refer to the blocked-event-client-y property of violation reports for a description of this property.
blockedEventType of type DOMString, readonly
Refer to the blocked-event-type property of violation reports for a description of this property.
blockedTargetID of type DOMString, readonly
Refer to the blocked-target-id property of violation reports for a description of this property.
blockedTargetXPath of type DOMString, readonly
Refer to the blocked-target-xpath property of violation reports for a description of this property.
deviceHeight of type long, readonly
Refer to the device-height property of violation reports for a description of this property.
deviceWidth of type long, readonly
Refer to the device-width property of violation reports for a description of this property.
pointerHeight of type long, readonly
Refer to the pointer-height property of violation reports for a description of this property.
pointerType of type DOMString, readonly
Refer to the pointer-type property of violation reports for a description of this property.
pointerWidth of type long, readonly
Refer to the pointer-width property of violation reports for a description of this property.
touchEvent of type bool, readonly
Refer to the touch-event property of violation reports for a description of htis property.
partial dictionary SecurityPolicyViolationEventInit {
    DOMString blockedEventType;
    bool      touchEvent;
    DOMString pointerType;
    long      pointerHeight;
    long      pointerWidth;
    long      deviceHeight;
    long      deviceWidth;
    long      blockedEventClientX;
    long      blockedEventClientY;
    DOMString blockedTargetID;
    DOMString blockedTargetXPath;
};

9.1.2 Dictionary SecurityPolicyViolationEventInit Members

blockedEventClientX of type long
Refer to the blocked-event-client-x property of violation reports for a description of this property.
blockedEventClientY of type long
Refer to the blocked-event-client-y property of violation reports for a description of this property.
blockedEventType of type DOMString
Refer to the document-uri property of violation reports for a description of this property.
blockedTargetID of type DOMString
Refer to the blocked-target-id property of violation reports for a description of this property.
blockedTargetXPath of type DOMString
Refer to the blocked-target-xpath property of violation reports for a description of this property.
deviceHeight of type long
Refer to the device-height property of violation reports for a description of this property.
deviceWidth of type long
Refer to the device-width property of violation reports for a description of this property.
pointerHeight of type long
Refer to the pointer-height property of violation reports for a description of this property.
pointerType of type DOMString
Refer to the pointer-type property of violation reports for a description of this property.
pointerWidth of type long
Refer to the pointer-width property of violation reports for a description of this property.
touchEvent of type bool
Refer to the touch-event property of violation reports for a description of htis property.

9.2 SecurityPolicy

Let the active CSP policies be the set of CSP policies the user agent is currently enforcing for the associated document.

partial interface Security {
    readonly    attribute bool inputProtection;
};

9.2.1 Attributes

inputProtection of type bool, readonly
A boolean representing the logical or of whether the input-protection directive is present or implied in each of the active CSP policies. [CSP11]

10. Input Protection Heuristic

This section is non-normative.

The algorithm described here can be implemented mostly in terms of HTML5 constructs, but requires the ability to monitor and intercept actions in the rendering of a resource and delivery of events to that resource. User agents may apply equivalent protections using means more optimized for their implementation details, may ignore recommendations where the browsing environment eliminates certain classes of attack, (e.g. the cursor sanity check in a touch-only environment) or may implement some features in terms of the underlying operating system or platform rather than directly in the user agent.

10.1 Preparation

  1. Listener registration - On the topmost window, register a "global" capturing event listener for mouse button, tapping, keyboard, drag & drop and focus events, which must be guaranteed to run before any other event handler of the same kind and therefore be able to prevent any event from being handled by the content, if needed.
  2. Display changes tracking - whenever a repaint occurs in the topmost window or in one of its descendants, create a record containing a weak reference to the Origin causing the repaint, the screen coordinates of the regions being repainted and a timestamp detailing when the repaint occurred, and add this record to a screen-global list named "Display Changes List". Records older than the maximum value for input-protection display-time can be discarded on update.

10.2 UI Event handling

  1. Timing attacks countermeasure - check whether the "Display Change List" contains any record younger than the input-protection display-time value, whose repainted regions intersect with the protected UI elements and whose repaint-causing Origin is different than the protected one. If this is true, hinting at a recent change in the way the protected UI is displayed, with causes external to the UI itself (e.g. an overlapping element in an ancestor document or a floating window being suddenly moved away), assume a timing attack is happening and jump to step 4.
  2. Cursor sanity check - By querying computed-style with the ":hover" pseudo-class on the element (if the target is plugin content) or on the host frame element and its ancestors (if the target is a nested document), check whether the cursor has been hidden or changed to an possibly attacker-provided bitmap: if it has, jump to step 4. This provides protection against "Phantom cursor" attacks, also known as "Cursorjacking".
  3. Obstruction check Take two screenshots of the area defined by the input-protection-clip and input-protection-selectors directives and containing the DOM element which is about to receive the event.
    1. The control image is taken from its owner document's "point of view" (unobstructed by definition) in an off-screen HTML5 canvas element [HTML5]. The user image is taken from either the topmost window's point of view in an off-screen HTML5 canvas element [HTML5] or using the fully compositied operating system perspective, obtained using OS-native APIs.
    2. When this heuristic is applied to plugin content, the control image must contain the element itself only.
    3. If the number of the pixels which are different between the screenshots don't exceed a percentage threshold defined by the tolerance property of the input-protection directive, return.
    4. Differences are computed at a pixel-by-pixel level. Any difference in the value of a pixel and it does not match. For example, a protected area in blue overlayed entirely by cross-origin content in red at 1% opacity is considered to be 100% different, not 1% different. If portions of the control image are clipped by the view port or otherwise occluded, all such pixels must be considered not to match.
    5. Otherwise, if the differences exceed the tolerance, assume that the DOM element which the user is interacting with has been obstructed or obscured by a UI Redressing attempt and proceed with step 4.
  4. Violation management - If in report-only mode, set the unsafe property of the event been handled to true and let the event processing continue. Otherwise, prevent the event from reaching its target. Create and send a violation report if a valid report-uri has been specified.
Note: Implementation Note

In the first implementation of this hueristic, NoScript's ClearClick, the screenshots are taken by using the CanvasRenderingContext2D.drawWindow() method, which is a Mozilla-proprietary extension of the HTML 5 Canvas API available to privileged code only, allowing the content of DOM windows to be drawn on a canvas surface exactly as rendered on the screen. The rest of this phase relies on cross-browser canvas features, instead, such as pixel grabbing and data URL serialization.

11. Alternate Heuristic

This section is non-normative.

Some user agents use a strategy for hit testing and delivering UI events involving multiple composited layers managed on a GPU. This alternative heuristic describes one possible implementation strategy for the input-protection directive in this architecture that may be a better fit than the standard heuristic.

GPU-optimized user agents typically separate the browser UI process from the process that handles building and displaying the visual representation of the resource. (In this context the term "process" refers to any encapsulated subunit of user-agent functionality that communicates to other subunits through message passing, without implying any particular implementation details such as locality to a thread, OS-level "processes" or the like.) It is typical for the browser UI process to receive user events such as mouse clicks and then marshal these to the render process, where the event is hit tested through the page's DOM, checking for event handlers along the way. As an optimization, the render process may communicate hit test rectangles back to the UI process in advance so that the UI process can immediately respond to, e.g. a Touch event by scrolling, if the event target falls within coordinates for which there are no other registered handlers in the DOM. A similar strategy can be used to create an implementation of the input protection heuristic that is consistent with this multi-process, compositing architecture.

If a resource is being loaded in a frame, iframe, object, embed or applet context specifies an input-protection directive, apply the following steps:

11.1 Preparation

  1. Protected hit test rectangle tracking: Hook the creation of event handlers for protected events and elements and add the DOM nodes with any such handler to a collection. After a layout occurs, or when an event handler is added or removed, iterate across all DOM nodes to generate a vector of rectangles and their associated Origins where events must be checked for safety. If the input-protection applies to the DOMWindow or Document node, avoid this expensive process of walking the renderers and simply use the view's bounds, as they're guaranteed to be inclusive.
  2. (Optionally) Put the protected areas into a backing store / composited layer: To avoid the expense of having to re-layout and re-paint protected regions during the obstruction check, it may make sense to designate and place these regions into their own backing store or composited layer which can serve as a cached control image.
  3. Display changes tracking: whenever a region in a protected hit test rectangle is invalidated, create a record containing a weak reference to the Origin causing the repaint, the screen coordinates of the regions being repainted and a timestamp detailing when the repaint occurred, and add this record to a screen-global list named "Display Changes List". Records older than the maximum value of input-protection display-time can be discarded on update.

11.2 UI Event handling

  1. Hit testing in the compositor: When an event is received, check whether it is on any layer and then walk the layer hierarchy checking the protected hit test rectangles on every layer. If there is a hit, continue this heuristic. Otherwise, exit this heuristic and event processing proceeds as normal.
  2. Timing attacks countermeasure check whether the "Display Change List" contains any record younger than the input-protection display-time value, whose repainted regions intersect with the protected regions and whose repaint-causing Origin is different than the protected one. If this is true, hinting at a recent change in the way the protected UI is displayed, with causes external to the UI itself (e.g. an overlapping element in an ancestor document or a floating window being suddenly moved away), assume a timing attack is happening and jump to Violation management.
  3. Cursor sanity check: By querying computed-style with the ":hover" pseudo-class on the element (if the target is plugin content) or on the host frame element and its ancestors (if the target is a nested document), check whether the cursor has been hidden or changed to a possibly attacker-provided bitmap. If it has, proceed to Violation management. This provides protection against "Phantom cursor" attacks, also known as "Cursorjacking".
  4. Obstruction check: Compare two sets of pixels: the control image is the protected region as if it was rendered alone, unobstructed by pixels originating from any other document context. If the protected regions were placed into their own backing store / composited layer, this should be readily available, although the pixels may need to be read back from the GPU to perform a comparision. The user image represents the same area as the control image in the outermost document's coordinate system and contains the final set of common pixels for the fully rendered page. The control image can be acquired through operating system APIs or from the compositor for the outermost document context. These images are compared, and if the number of pixels that differ are below the tolerance threshold associated with the input-protection directive, proceed to deliver the event normally, otherwise proceed to Violation management.

    Differences are computed at a pixel-by-pixel level. Any difference in the value of a pixel and it does not match. For example, a protected area in blue overlayed entirely by cross-origin content in red at 1% opacity is considered to be 100% different, not 1% different. If portions of the control image are clipped by the view port or otherwise occluded, all such pixels must be considered not to match.

    As a short-cut, a user agent MAY choose to treat any pixels in a protected layer with an opacity of less than 100% as failing to match by definition. In cases where a fully-composited user view is not available or extremely expensive to calculate, this optimization allows the obstruction check to be performed with only a knowledge of the layers that fall on top of the protected layer.

  5. Violation management - If in report-only mode, set the unsafe property of the event been handled to true and let the event processing continue. Otherwise, prevent the event from reaching its target. Create and send a violation report if a valid report-uri has been specified.
Note: Implementation note

Optimized and potentially cross-platform implementations of screen and cursor capturing and monitoring regions for invalidation may be available as part of e.g. screen- sharing functionality through getUserMedia() [MEDIACAPTURE] or other remote desktop-type functionality available in certian user agents, e.g. the ScreenCapturer interface in Chromium.

12. Examples

12.1 Sample Policy Definitions

This section is non-normative.

This section provides some sample use cases and accompanying security policies.

A resource wishes to block delivery of UI events to the document unless its whole body has been entirely visible (no tolerance) during the past 1 second (default display-time value):

Example 3: Policy Header
Content-Security-Policy: input-protection

A resource wishes to block delivery of UI events to the element with id "send-box", all the elements with class ".tweet" and all the forms in the page unless those elements have been visible for the past 800 milliseconds at least, (their intrinsic sizes is used as a reference for screenshot comparison):

Example 4: Policy Header
Content-Security-Policy: input-protection display-time 800;
        input-protection-selectors #send-button, .tweet, form

A resource wishes to block delivery of UI events to any obstructed HTML button and suggests a 15% tolerance threshold for determining obstruction of the element with a 200 pixels wide margin above and before (on the top and on the left, if orientation is LTR) the triggering element:

Example 5: Policy Header
Content-Security-Policy: input-protection tolerance=15;
                input-protection-selectors above=200 before=200 after=0 below=0 button, input[type=submit], input[type=button]

A resource wishes to receive reports when the UI Security heuristic is triggered for any element in the <body>, with the default 300 by 300 pixels clipped reference area and 0 tolerance:

Example 6: Policy Header
Content-Security-Policy-Report-Only: input-protection; input-protection-clip;
                                     report-uri https://example.com/csp-report?unique_id=XKSJ9KAAHJDK9928KKSJEQ

A resource wants to react to potential clickjacking directly, without sending a report, so it sets a report-only header but does not specify a report-uri. When a UIEvent is sent, the unsafe attribute will still be set when the heuristic is triggered:

Example 7: Policy Header
Content-Security-Policy-Report-Only: input-protection

A resource wants to allow itself to be embedded by ancestors that are same-origin or from the origin https://checkout.example.com, but also to have the unsafe attribute set on events that violate the input protection heuristic.

Example 8: Policy Header
Content-Security-Policy: frame-options 'self' https://checkout.example.com
Content-Security-Policy-Report-Only: input-protection 

12.2 Sample Violation Report

This section is non-normative.

This section contains an example violation report the user agent might sent to a server when the protected resource violations a sample policy.

In the following example, a document from http://example.org/page.html was rendered with the following CSP policy:

input-protection; report-uri https://example.org/csp-report.cgi?unique_id=12345

A click violated the policy.

Example 9: Sample violation report JSON body
{
  "csp-report": {
    "document-uri": "http://example.org/page.html",
    "referrer": "http://evil.example.com/haxor.html",
    "blocked-event-type": "click",
    "blocked-event-client-x": "325",
    "blocked-event-client-y": "122",
    "touch-event": "false",
    "device-width": "800",
    "device-height": "300",
    "blocked-target-xpath": "/html[0]/body[0]/div[6]/form[2]/input[0]",
    "violated-directive": "input-protection",
    "original-policy": "input-protection; report-uri https://example.org/csp-report.cgi?unique_id=12345"
  }
}

12.3 Example Boundary Calculations for the Obstruction Check

A resource at OriginX embeds a resource at OriginY. The OriginY resource has the following policy:

Content-Security-Policy: input-protection tolerance=50; input-protection-selectors div;

and results in the following layout:

Example frame layout.

The element with the id "div1" has an onClick handler defined, and a click event is triggered at 120,120 in the OriginX document's coordinate system. The red dot indicates the position of the event. The event is delivered to "div1", which matches the input-protection-selectors, and no parent of "div1" matches. As no input-protection-clip value is defined, the entire area of "div1" becomes the boundaries for the obstruction check, indicated by the cyan fill. As more than 50% of this area is occluded behind the iframe viewport, and so does not match by definition, this will trigger a violation.

Example frame layout showing selector policy with an event.

If the OriginY protected resource set the following policy, instead:

Content-Security-Policy: input-protection tolerance=50; input-protection-selectors div; input-protection-clip before=60 after=60 above=60 below=60;

The region for the obstruction check, still indicated in solid cyan, is now only the intersection of the boundaries of the protected element handling the event, indicated by diagonal cyan lines, and the clipping window around the event, indicated by the green dotted line. If the OriginX resource has not painted anything over the iframe viewport, this check will not trigger a violation because the entire cyan area will be identical in the user image and control image.

Example frame layout showing selector policy with an event.

If the OriginY protected resource omitted selectors, as in this policy:

Content-Security-Policy: input-protection tolerance=50; input-protection-clip before=60 after=60 above=60 below=60;

The region for the obstruction check, still indicated in solid cyan, is now the intersection of the boundaries of the entire document, indicated by diagonal cyan lines, and the clipping window around the event, indicated by the green dotted line. This demonstrates that portions of the protected resource may be included in the obstruction check region, even if they do not have event listeners. Thus, the hit test rectangles which trigger the heuristic do not necessarily compose the entire region that must be checked.

As in the previous example, if the OriginX resource has not painted anything over the iframe viewport, this check will not trigger a violation because the entire cyan area will be identical in the user image and control image.

Example frame layout showing selector policy with an event.

13. Security Considerations

This section is non-normative.

UI Redressing and Clickjacking attacks rely on violating the contextual and temporal integrity of embedded content. Because these attacks target the subjective perception of the user and not well-defined security boundaries, the heuristic protections afforded by the input-protection directive can never be 100% effective for every interface. It provides no protection against certain classes of attacks, such as displaying content around an embedded resource that appears to extend a trusted dialog but provides misleading information.

14. Implementation Considerations

This section is non-normative.

The policy and intent of the user always takes precedence over the policy of resources. In particular, transformations, customizations or enhancements of visual content made by the user agent or user-installed plugins SHOULD NOT cause the input-protection heuristic to be triggered.

Many UI Redressing and Clickjacking attacks rely on exploiting specific features of user agents, such as repositioning of the browsing window, hiding or creating fake cursors, and script-driven scrolling and content repositioning. Not all attacks apply to all user agents in all contexts. User agents are free to optimize or not implement suggested heuristics when they do not apply, for example:

Some resource owners may specify a restrictive policy forbidding embedding in user agents that only understand X-Frame-Options but be more permissive with user agents that implement UI Security directives. User agents that are aware of but choose not to implement any of the heuristics in this document MAY still ignore X-Frame-Options when presented in combination with UI Security directives in a Content Security Policy. For example, a browsing environment that deliberately chooses not to implement UI Security features because they interfere with assistive technologies SHOULD NOT deny users access to resources on this account. User agents taking this stance SHOULD implement the unsafe attribute of the UIEvent interface as this may be interrogated by client applications doing feature detection.

In environments that support multiple, overlapping browser windows, attacks may be mounted by positioning a target window under another, instructing the user to double click, and closing the obstructing window with the first click. [CLICKJACKING-Unresolved] In such environments user agent implementers may wish to use a native operating system screenshot facility to calculate the user's view for the obstruction check phase of the heuristic. In such cases user agents should take special caution to potential infereference from accessibility technologies

While this document describes a mechanism for resource authors to opt-in to User Interface Security protections, user agents MAY choose to opt-in resources to input-protection by default, or provide users with an option to manually enable such protections.

If a user agent or user chooses to apply input protection in the absence of an explicit directive, violations SHOULD NOT cause a violation report to be generated, even if the resource supplied a Content Security Policy with a report-uri.

14.1 Accessibility Technologies

This section is non-normative.

Certain classes of accessibility technologies such as screen readers will provide strong defenses against many classes of UI Redressing attacks by presenting the content to the user in a manner not subject to interference. Such user agents SHOULD set the unsafe attribute of the UIEvent interface as this may be interrogated by client applications doing feature detection, and SHOULD ignore X-Frame-Options headers when presented in combination with UI Security directives in a Content Security Policy.

Use of accessibility technologies MUST NOT by itself cause the input-protection heuristic to be triggered. Accessibility technologies that modify the appearance of a resource, such as screen magnifiers or color and contrast modifications to the display have the potential to interfere with the obstruction checkif not applied in a consistent manner to both the user image and control image. To prevent this inteference, user agents SHOULD apply accessibility transformations to the control image if possible. If a user agent is able to detect that accessibility technologies are in use that cannot be applied uniformly as part of the obstruction check, the check MUST be disabled. In some cases, interference from accessiblity tools may be avoided by acquiring the user image in terms of the user agent's local rendering surface, rather than using an operating-system level screenshot.

User agents SHOULD provide a means for the user to manually disable enforcement of the Input Protection Heuristic if it interferes with their chosen accessibility technologies.

15. Implementation Considerations for Resource Authors

This section is non-normative.

When possible, resource authors SHOULD make use of violation reports and the unsafe attribute to apply additional security measures in the application or during back-end processing. Real-time measures in the application might include requiring completion of a CAPTCHA [CAPTCHA-Wikipedia] or responding to an out-of-band confirmation when the UI Security heuristic is triggered. Example back-end measures might include increasing a fraud risk score for individual actions that trigger or targets accounts/resources that frequently trigger UI Security heuristics. To be able to do this effectively, it is likely necessary to encode into the report-uri a unique identifier that can be correlated to the authenticated user and the action they are taking.

16. IANA Considerations

This document does not define new message headers and uses the existing grammar of the Content-Security-Policy and Content-Security-Policy-Report-Only headers, so no updates to the permanent message header field registry (see [RFC3864]) are required.

A. References

A.1 Normative references

[ABNF]
D. Crocker; P. Overell. Augmented BNF for Syntax Specifications: ABNF. January 2008. STD. URL: http://www.ietf.org/rfc/rfc5234.txt
[CSP]
B. Sterne and A. Barth Content Security Policy 1.0. W3C Candidate Recommendation. (Work in progress.) URL: http://www.w3.org/TR/2012/CR-CSP-20121115/
[CSS3-MEDIAQUERIES]
Håkon Wium Lie; Tantek Çelik; Daniel Glazman; Anne van Kesteren. Media Queries. 19 June 2012. W3C Recommendation. URL: http://www.w3.org/TR/css3-mediaqueries
[DOM-LEVEL-2-CORE]
Arnaud Le Hors; Philippe Le Hégaret; Lauren Wood; Gavin Nicol; Jonathan Robie; Mike Champion; Steven B Byrne et al. Document Object Model (DOM) Level 2 Core Specification. 13 November 2000. W3C Recommendation. URL: http://www.w3.org/TR/DOM-Level-2-Core/
[DOM-LEVEL-2-EVENTS]
Tom Pixley. Document Object Model (DOM) Level 2 Events Specification. 13 November 2000. W3C Recommendation. URL: http://www.w3.org/TR/DOM-Level-2-Events/
[HTML401]
Dave Raggett; Arnaud Le Hors; Ian Jacobs. HTML 4.01 Specification. 24 December 1999. W3C Recommendation. URL: http://www.w3.org/TR/html401
[HTML5]
Robin Berjon; Steve Faulkner; Travis Leithead; Erika Doyle Navara; Edward O'Connor; Silvia Pfeiffer. HTML5. 4 February 2014. W3C Candidate Recommendation. URL: http://www.w3.org/TR/html5/
[ORIGIN]
A. Barth. The Web Origin Concept. December 2011. RFC. URL: http://www.ietf.org/rfc/rfc6454.txt
[RFC2119]
S. Bradner. Key words for use in RFCs to Indicate Requirement Levels. March 1997. Internet RFC 2119. URL: http://www.ietf.org/rfc/rfc2119.txt
[RFC7034]
D. Ross and T. Gondrom, IETF HTTP Header X-Frame-Options. Internet RFC 7034 URL: http://tools.ietf.org/html/rfc7034
[TOUCH-EVENTS]
Doug Schepers; Sangwhan Moon; Matt Brubeck; Arthur Barstow. Touch Events. 10 October 2013. W3C Recommendation. URL: http://www.w3.org/TR/touch-events/
[URI]
T. Berners-Lee; R. Fielding; L. Masinter. Uniform Resource Identifiers (URI): generic syntax. January 2005. RFC 3986. URL: http://www.ietf.org/rfc/rfc3986.txt
[XPATH]
James Clark; Steven DeRose. XML Path Language (XPath) Version 1.0. 16 November 1999. W3C Recommendation. URL: http://www.w3.org/TR/xpath

A.2 Informative references

[CAPTCHA-Wikipedia]
Wikipedia CAPTCHA from Wikipedia. URL: http://en.wikipedia.org/wiki/CAPTCHA
[CLEARCLICK]
G. Maone ClearClick: Effective Client-Side Protection Against UI Redressing Attacks. (Work in progress.) URL: http://noscript.net/downloads/ClearClick_WAS2012_rv2.pdf
[CLICKJACKING-Unresolved]
Lin-Shung Huang and Collin Jackson. Clickjacking Attacks Unresolved. Carnegie Mellon University, 06 July 2011. URL: https://docs.google.com/document/pub?id=1hVcxPeCidZrM5acFH9ZoTYzg1D0VjkG3BDW_oUdn5qc
[CSP11]
A. Barth, D. Veditz and M. West Content Security Policy 1.1. W3C Editors' Draft. (Work in progress.) URL: https://dvcs.w3.org/hg/content-security-policy/raw-file/tip/csp-specification.dev.html
[FRAMEBUSTING]
Boneh, et al. Busting frame busting: a study of clickjacking vulnerabilities at popular sites. URL: http://seclab.stanford.edu/websec/framebusting/
[INCONTEXT]
Lin-Shung Huang, et al. Clickjacking:Attacks and Defenses published in the 21st USENIX Security Symposium Proceedings. URL: https://www.usenix.org/system/files/conference/usenixsecurity12/sec12-final39.pdf
[MEDIACAPTURE]
D. Burnett, A. Bergkvist, C. Jennings and A. Narayanan Media Capture and Streams. W3C Working Draft (Work in progress.) URL: http://www.w3.org/TR/mediacapture-streams/
[POINTER-EVENTS]
Jacob Rossi and Matt Brubeck. Pointer Events. 19 February 2013 W3C Working Draft. (Work in progress.) URL: http://www.w3.org/TR/pointerevents/
[SELECTORS4]
Elika J. Etemad. Selectors Level 4. 29 September 2011. W3C Working Draft. (Work in progress.) URL: http://www.w3.org/TR/2011/WD-selectors4-20110929/
[UIREDRESS]
M. Zalewski Browser Security Handbook, part 2. URL: http://code.google.com/p/browsersec/wiki/Part2#Arbitrary_page_mashups_(UI_redressing)