5 ms·
Any ideas why this is branded as "Docker"? Are the same namespacing constructs not being used by other Linux container runtimes? I think this should be titled "
by alpb 9y ago
Any ideas why this is branded as "Docker"? Are the same namespacing constructs not being used by other Linux container runtimes? I think this should be titled "Escaping Linux containers" as docker is not at fault here?
- derefr 9y ago"Docker" isn't at fault here either way, as Docker isn't its own "execution driver" (in Docker parlance) any more; that would be https://github.com/opencontainers/runc https://github.com/opencontainers/runc. But to answer the spirit your question, each container runtime uses its own peculiar combination of such constructs. It's helpful to know that this attack allows you to break out of the combination used specifically by runc, and thereby to break out of any system relying on Docker (with the default runc execution driver.)
- icebraining 9y agoI doubt any runtime would have prevented this bug; some have seccomp-bpf profiles to blacklist some kernel operations, but I doubt any block a function as basic as waitid().
- jo909 9y agoWhat do you mean by "branded"? The author shows a concrete exploit of the kernel bug described in CVE-2017-5123 as he has developed it in the context of the docker container environment. He shows how to use this bug to break out of docker, so he calls the blog post "Escaping Docker ...". Which is IMHO the most interesting container runtime to write such an exploit for first because it is very widely deployed, but it might also just have been what the author is most familiar with or what was easiest to develop for him.
- chowyuncat 9y agoThink of it this way: what if the author had titled it "Escaping Ubuntu containers" ?
- dchest 9y agoWhy? The article demonstrates exploitation of Docker containers.
- oblio 9y agoYeah, but is it limited to Docker containers? Can other container types be attacked in the same way?
- dchest 9y agoSure, you can write an article demonstrating exploitation of Ubuntu containers and call it "Escaping Ubuntu containers".
- chowyuncat 9y agoThe former part was exactly what was done.
- jo909 9y agoIt is a reasonable _assumption_ that other container runtimes on linux might be affected by the same kernel bug. The article does not explore that and the author has no duty to do so just to avoid using a branded technology name. How would you reasonably talk about "Linux containers" without having a very exhaustive list of all existing implementations and testing all of them? If one of them is not affected you are now factually wrong.
- chowyuncat 9y agoThe exploit overwrites kernel memory credentials of a task structure. That structure is the lynchpin of kernel security, including SELinux.
- jo909 9y ago"Ubuntu container" is just not a name typically used for anything. "Docker container" is. Let's make it realistic and say he had used RedHat OpenShift as his target and example for the exploit. I'd be completely fine with the title referencing that exact product by name. Why would he have to dance around what he is using in his demo? Maybe that concrete product has multiple layers of security or lacks them, or uses a certain version etc. He can only speak to what he himself was using and testing. "Escaping Docker container..." is the best short description (as you would need it for a title) of this demo exploit I can think of.
- shykes 9y agoMarketing. There's a long tradition by enterprise vendors large and small to market someone else's product as insecure, in order to create demand for their "improved, secured" version. In this particular instance, Twistlock is selling Docker security by amplifying the meme of "insecure Docker". The Docker brand has visibility the target audience (Enterprise IT) so it's a good target for this kind of piggyback. This type of FUD marketing happens all the time in many different markets, it's not specific to Docker.