3 ms·
That is not necessarily sufficient, because there are many, many variations on this attack. For example, suppose my target site has an asset at a well known UR
by patio11 16y ago
That is not necessarily sufficient, because there are many, many variations on this attack.
For example, suppose my target site has an asset at a well known URL, such as:
http://www.google.com/intl/en_ALL/images/srpr/logo1w.png http://www.google.com/intl/en_ALL/images/srpr/logo1w.png
I could execute Javascript on your machine to load that in a standard IMG tag, in a div with display: none set on it. I could then, via a variety of fairly unobtrusive methods, time how long it takes your browser to load that image.
If you have been to Google recently, that image will be cached, and your browser will load it almost instantaneously. If you have not been to Google recently, that image will not be in your cache, and my Javascript will trivially be able to tell the difference between "presence of a network request going through the Internet" and "absence of said request."
You don't even really need the asset if you want to be tricky about it. (I can time your DNS cache, too, it just takes a little more work.)
- detst 16y agoI was thinking recently about how a similar technique could be used on DropBox. Government employee uploads "sensitive" data and sees that no actual data is transfered. They now have a person of interest in a leak or possession case.
- bdonlan 16y agoThat requires the data the government employee has to be byte-for-byte identical to the one already in dropbox, which is unlikely for an unintended leak.
- jacquesm 16y agoTrue, it's not a fix to all attacks or even a fix for all variations of this attack but it does close this particular hole. For more interesting stuff see: http://browserspy.dk/ http://browserspy.dk/ More than you ever wanted to know about how your browser leaks data.