TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Sidechannel pixel-stealing attack works in Chromium on all modern GPUs

209 pointsby anfiltover 1 year ago

12 comments

vessenesover 1 year ago
This is a solid attack. I wouldn&#x27;t call it beautiful, it&#x27;s more like a well-considered thorough engineering tour-de-force. I&#x27;m horrified but applaud the team.<p>Here&#x27;s how it works: a stack of SVG filters is created. These filters are constructed so that they will tend to be faster processing a dark pixel than they will be processing a light pixel.<p>An iframe is loaded up by the attacking site, pointing at, say, a banking site or some other target of interest. I couldn&#x27;t find exact details, and my HTML skills are rusty, but I assume the iframe is a 1x1 pixel iframe, and given a pixel offset.<p>The SVG stack is loaded onto the iframe using CSS, then unloaded, then loaded, etc. a whole bunch of times, and average timing results are assessed. Based on these average times, the pixel is marked &#x27;dark&#x27; or &#x27;light&#x27;. Repeat for each pixel.<p>Average time per pixel is in the 1-2 second range. Per pixel. So, this is a slow attack. It could probably get an order of magnitude faster with a sort of combo of zooming and greyscale heuristics that resolves over time, though.<p>They have a number of cool graphs showing the broad spread of times, and it does look easy to distinguish; their success varies by architecture, but it&#x27;s over 96% for almost every architecture they test. They show it works while multiple videos and other things that tax the GPU are playing.<p>Proposed fix: let browsers tell the GPU they need some variant of constant-time processing for an iframe. Which is super, super gross.<p>Safari and Firefox don&#x27;t currently allow cross-site iframe injection, so the attack only works on Chromium-line browsers.<p>Again, eww. And, wow!
评论 #37666765 未加载
评论 #37666560 未加载
评论 #37668672 未加载
stuaxoover 1 year ago
Has anyone made a law about this sort of thing yet? If not I&#x27;ll have it:<p>Axons Law - Any sufficiently fast optimisation will be repurposed as an attack vector.
评论 #37667470 未加载
评论 #37675953 未加载
评论 #37672709 未加载
anfiltover 1 year ago
The fact only chromium based browsers are effected makes me think really the problem lies with the browser. However, still using the result of compression as a delta to extract sensitive data is not exactly new.
评论 #37664222 未加载
pestsover 1 year ago
Pixel by pixel.<p>&gt; On AMD’s Ryzen 7 4800U, GPU.zip took about 30 minutes to render the targeted pixels with 97 percent accuracy. The attack required 215 minutes to reconstruct the pixels when displayed on a system running an Intel i7-8700.
zorgmonkeyover 1 year ago
PoC code: <a href="https:&#x2F;&#x2F;github.com&#x2F;UT-Security&#x2F;gpu-zip">https:&#x2F;&#x2F;github.com&#x2F;UT-Security&#x2F;gpu-zip</a><p>Preprint paper: <a href="https:&#x2F;&#x2F;www.hertzbleed.com&#x2F;gpu.zip&#x2F;GPU-zip.pdf" rel="nofollow noreferrer">https:&#x2F;&#x2F;www.hertzbleed.com&#x2F;gpu.zip&#x2F;GPU-zip.pdf</a>
Dweditover 1 year ago
uMatrix-like extensions will stop this, making you explicitly allow the iFrame to run before it can happen.
评论 #37672408 未加载
hakreover 1 year ago
Isn&#x27;t using GPU and hardware acceleration considered experimental in browsers and the safe default to disable such features for day-to-day use?<p>Not saying to make this achievement smaller than it is, quite the opposite, it&#x27;s important there is more such research.
评论 #37671105 未加载
quinncomover 1 year ago
Until Google decides whether and how to patch, it seems blocking all third-party iframes can mitigate this issue, using this uBlock Origin rule:<p>||*^$subdocument,third-party
tedunangstover 1 year ago
But is it really stealing if the owner isn&#x27;t deprived of use? It&#x27;s just costless duplication of a nonscarce resource.
评论 #37665604 未加载
评论 #37665267 未加载
评论 #37665356 未加载
评论 #37665753 未加载
评论 #37670882 未加载
评论 #37666038 未加载
评论 #37665458 未加载
chromoblobover 1 year ago
Very good. A step towards the proper concept of computer security.
pshcover 1 year ago
Could we, like, do away with iframes?
评论 #37666375 未加载
评论 #37666450 未加载
评论 #37670651 未加载
评论 #37669072 未加载
baz00over 1 year ago
We should have never gone past Gopher :)
评论 #37666463 未加载