<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none"><!-- p { margin-top: 0px; margin-bottom: 0px; }--></style>
</head>
<body dir="ltr">
<div id="OWAFontStyleDivID" style="font-size:12pt;color:#000000;background-color:#FFFFFF;font-family:Calibri,Arial,Helvetica,sans-serif;">
<style type="text/css" style="">
<!--
p
{margin-top:0px;
margin-bottom:0px}
-->
</style>
<div id="OWAFontStyleDivID" style="font-size:12pt; color:#000000; background-color:#FFFFFF; font-family:Calibri,Arial,Helvetica,sans-serif">
<p></p>
<div style="color:rgb(33,33,33)">
<div>
<div dir="ltr">
<div>Hi Fabrice,</div>
<div><br>
</div>
<div>> Hi all,</div>
<div>> This topic came up in my previous one about window placement, and I'd like to</div>
<div>> go further.</div>
<div>> So currently there is no such thing as Global shortkeys and keyboard focus,</div>
<div>> however let me present a typical real use-case:</div>
<div>> [...]</div>
<div>> </div>
<div>> Now, I've read some vague things about privileged clients, is it still being </div>
<div>> considered ?</div>
<div><br>
</div>
<div>Note that most of the formalisation that occurred was done by Martin Peres from</div>
<div>Nouveau (and to a lesser extent myself; I'm not a Wayland dev at all though).</div>
<div><br>
</div>
<div>We're hoping to have a number of PoC demos of privileged clients for XDC but are</div>
<div>both really really busy with our research (both PhD students). It's not clear if</div>
<div>we will have implemented stuff to demonstrate... If you want/need</div>
<div>to work on global shortcuts in the next weeks I can make an effort to make our</div>
<div>latest discussions and plans available in a concise form.</div>
<div><br>
</div>
<div>> Would it be some Android-like capabilities that the user validates on </div>
<div>> installation or the first time they are required by the application ?</div>
<div>> What are the plans for these 2 key features ?</div>
<div><br>
</div>
<div>We only discussed what the privileges are. Intercepting global shortcuts is a</div>
<div>privilege so your app would need to either:</div>
<div>- have a capability to register a global shortcut itself</div>
<div>- be entitled by a trusted third party to using a specific global shortcut</div>
<div><br>
</div>
<div>A capability should be granted to a package by a distributor, most likely. This</div>
<div>means distros who care about security would setup a process to verify why app</div>
<div>devs/packagers want a capability for their app (whilst allowing core projects</div>
<div>such as DEs/distro apps to have privileges and be deployable right away).</div>
<div><br>
</div>
<div>The second point is a bit fuzzier, especially for global shortcuts. For some</div>
<div>privileged interfaces once apps can be sandboxed on Linux, [and once I've</div>
<div>written a decently secure UI embedding protocol *], they can be given widgets</div>
<div>from a trusted third-party that the user can interact with to organically grant</div>
<div>privileges. Apps should also have some nice APIs for opening and exporting</div>
<div>resources in a secure way.</div>
<div><br>
</div>
<div>You can tell that it's hard to find out how to provide a global shortcut UI</div>
<div>abstraction that is unambiguous to users, especially since I understand your app</div>
<div>will be GUI-less. Xfce has a GUI for assigning global shortcuts to commands, and</div>
<div>I believe other DEs do as well. This utility will typically be the one holding a</div>
<div>capability for intercepting any global shortcut.</div>
<div><br>
</div>
<div>Your app should normally not qualify for such a privilege, so make your event</div>
<div>triggerable via a CLI call and get users to assign the shortcut to your app. If</div>
<div>DEs are willing to grant you full global shortcut privileges without assessing</div>
<div>who you are, what your app does and in what ways your app can be compromised,</div>
<div>they will probably have security issues in the future.</div>
<div><br>
</div>
<div>Feel free to work with distrubotrs to sketch out a process for granting and</div>
<div>revoking capabilities to third-party apps, etc. but I think this problem goes</div>
<div>well beyond the scope of Wayland privileged interfaces!</div>
<div><br>
</div>
<div>PS: you were the person proposing to let apps know or adjust their position on</div>
<div>the screen? This, typically, creates vulnerabilities and makes trusted UI</div>
<div>embedding much harder if not compromised. If you have specific use-cases that</div>
<div>need to be supported, please come discuss them with us (#wayland-security on</div>
<div>Freenode or this ML, I guess) so we can think of secure ways to support your</div>
<div>needs without compromising the separation between clients and trusted UIs.</div>
<div><br>
</div>
<div>Thanks,</div>
<div>--</div>
<div>Steve Dodier-Lazaro</div>
<div>PhD student in Information Security</div>
<div>University College London</div>
<div>Dept. of Computer Science</div>
<div>Malet Place Engineering, 6.07</div>
<div>Gower Street, London WC1E 6BT</div>
<div>OpenPGP : 1B6B1670<br>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</body>
</html>