I have a hard time understanding why Red Hat patented the concept behind the idmapped mount work I did? I mean just overall why even do this and also why hasn't anyone ever talked to us about this? I don't care about patents but this is really weird given that this patent is given to a single person but none of the actual people involved in this are even mentioned. Then again, I know nothing about patents. 🤷
https://patents.justia.com/patent/11797357

#linux #kernel

@brauner As an ex-Red Hatter, I would just recommend you to reach out to some of their legal teams and ask.

My general experience working there many years is that they generally don't intend to harm anyone, but to preserve and ensure open source projects can continue to prosper without risking threats by less nicer corporations willing to go to courts to get control of your stuff as cheap as possible. Red Hat's mantra is to support open source projects first of all. But patents and intellectual property stuff is confusing stuff for us not being in the legal business.

Perhaps @richardfontana can help you get in touch with the right people inside Red Hat?

@dazo @brauner @richardfontana (Also ex-RHer here) Patents normally don't come from the 'company' they normally come from an individual engineer; so it's probably they just didn't know about what you had done and none of the people who reviewed it did either. It's a big company, and lots of corners reinvent stuff other people did without realising history.

@brauner You might already know this but IBM (Red Hat) is part of the Open Invention Network, a defensive patent pool for Linux distros. The idea is that "anyone can freely use these patents on Linux unless they sue us for patent infringement". If anyone gets sued, the entire patent pool of all organisations (including Microsoft) will be used to fight back.

One explanation might be that they want to bolster that patent pool so it remains a threat to adversaries.

@mesebrec @brauner Still, this is very strange. IBM (RedHat) can just claim the patent, for this "invention"? Would the same hold true for any other commercial enterprise? At the very least I would expect the explicit agreement of the actual "inventor/creator". Or is there some general agreement between RedHat and the Linux Foundation?

@mesebrec @brauner It's still stealing from the creator, regardless of any "good intentions".

@goebbe @brauner I thought Christian's question was "why is my previous employer patenting my open source work", instead of "why is an unrelated company patenting my open source work"

I assumed Christian worked for Red Hat or IBM while he did that work, but this does not seem to be the case.

@mesebrec @goebbe I never worked for Red Hat. The thing is that a bunch of people conceived this idea as far back as 2016 and we shipped a prior version of this with Ubuntu in 2016/17 and then I did an upstream implementation in 2019 where that patent was first claimed.

@brauner @mesebrec @goebbe It's just an unfortunate side effect of the way the patent office works. If they'd found things like the presentations about shiftfs from 2017, the examiner would likely have disallowed the patent. However, the examiner's search strategy (according to the file wrapper) was to type "user namespace host namspace container" into google scholar and cite everything that came back. It didn't turn up any of the source code or existing works.

@brauner Red Hat doesn't just patent random stuff. It's an engineer at Red Hat who would have done it, I think their name is on the patent. So maybe ask them directly? If someone at RH is parenting something they didn't invent then it would not usually reflect well on them.

@airlied @brauner well yes @giuseppe who is listed as the inventor of the patent.

@juliank @airlied @brauner @giuseppe You can get it from the file wrapper details:

https://patentcenter.uspto.gov/applications/16393811?application=

@jejb @juliank @airlied @brauner @giuseppe sometimes people just have the same ideas. you get sensible software engineers looking at container technology and thinking "hmm what might be useful", they're going to come up with a lot of the same thoughts. (this is one of the dumb things about software patents in general, after all). before anyone goes nuclear, it'd probably be best just to wait for the author of the patent to explain.

@adamw @jejb @juliank @airlied @brauner @giuseppe Prefixing everything I say here with "I'm not a lawyer or legally trained person" and "I'm biased". People converge on the same ideas all the time, it's natural, not saying that's the case here because I wasn't involved. But it generally benefit's the open-source community as a whole when Red Hat patents things. Red Hat's stance towards patents is to defend the open source community as well as Red Hat.

https://www.redhat.com/en/about/patent-promise

@adamw @juliank @airlied @brauner @giuseppe I think it's just a reflection of the sad fact that the orchestration container community (Docker, Rocket, Kubernetes, Openshift) have never really communicated well with the Kernel container community or even looked much at what each other is doing.

@brauner interesting but not too surprising. like e.g. for almost any possible way you could ever possibly think of exchanging key material between peers or groups, there's some company with a patent. i'd expect that if i have anything with at minimum two code blocks, there's an entity with a patent for that.

there's a place for software patents, when it is like an actual invention but for most part they're nonsense. good example of such might be marching cubes algorithm for instance, for which autodesk held patent up until 2005 (imho rightfully so)...