Microsoft & Others Form The eBPF Foundation

[rabcor]

So... Is this actually supposed to do something? In-kernel virtual machines I heard, I mean that sounds kinda interesting, creepy as all hell and largely defeating the point of VMs in the first place, but interesting nontheless..

I just don't see why we would want something like that in the mainline linux kernel, I don't see why this stuff shouldn't be out-of-tree or in a different kernel line targeted at whatever people actually want to use this feature.

This is clearly not a core feature that belongs in every linux kernel, reeks of pointless bloat to me.

All this work to make the kernel bigger when we should be working to make it smaller.

[jacob]

Can you imagine sitting at the linux "Foundation" meeting:

Microsoft: Ok guys you know if they run linux we can't control them,
we can't bend them over and mine their data.

Crowd(Google etc): No... no... why ?
Microsoft: Silence. I have a plan, we all joined this foundation
so we could corrupt linux and be able to control it, no ?
If the linux "Foundation" changes linux then who would stop us ?

CCP: Son, we have our own list of patches that you must apply and
if you don't apply them then no business in our country for you.

Phoronix readers: Hey guys, you know the latest conspiracy theory. They are out to get us. No passaran, prepare your tinfoil hats! Them evil companie$ shall not make my irrelevant toy OS into something usable in the real world! Them won't submit a single line of codez into the system I use to run Xlogo! Not a single bit in my systemz will come from Micro$oft or Google and of course when I need to get something real done, I'll use Winblob$ or Mac because my uber anti kapitalist Linux can't do it.

Right?

[timofonic]

Is eBPF a really good idea in long term cases? Many very experienced people disagree.

Those very big corps have lots of resources that make them possible to have LOTS of security layers, many of them extremely secret and NEVER will be published or known outside their extremely paranoid environments. Despite of that, they have LOTS of security leaks (is Google less prone than the rest?).

It may make certain stuff faster, such as load balancing and even replace iptables. What's the real cost of it?

Buzzwords aren't enough...

[onlyLinuxLuvUBack]

Can you imagine sitting at the linux "Foundation" meeting:

Microsoft: Ok guys you know if they run linux we can't control them,
we can't bend them over and mine their data.

Crowd(Google etc): No... no... why ?
Microsoft: Silence. I have a plan, we all joined this foundation
so we could corrupt linux and be able to control it, no ?
If the linux "Foundation" changes linux then who would stop us ?

CCP: Son, we have our own list of patches that you must apply and
if you don't apply them then no business in our country for you.

[onlyLinuxLuvUBack]

Enqueue the Microsoft hatred

Looking at the list of evil companies, they would all like to control your life,content,freedom and your os.

[debrouxl]

Ten years from now when security researchers find all the security holes, there are going to be some painful mitigations for dealing with this.

Exactly. They're already working on it:

Thread https://mobile.twitter.com/andreyknv...97975979102217 : "The cool part about eBPF-based rootkits is portability.
A kernel module–based rootkit needs to be rebuilt when a new kernel is deployed."

Fortunately, Debian has recently started disabling unprivileged BPF by default on the sid kernels.

[GruenSein]

The more I hear about features implemented in Linux kernel, the more I think about Hurd. I mean, a lot of those things implemented in kernel, sound like some userspace thing.

Wouldn't that make HURD the opposite? AFAIK, HURD aims to provide most functionality as independent servers so that it is independent from other functionality and cannot crash the kernel itself. Lately, most Linux news seems to revolve around putting more stuff into the kernel.

[andyprough]

Ten years from now when security researchers find all the security holes, there are going to be some painful mitigations for dealing with this.

[etam]

The more I hear about features implemented in Linux kernel, the more I think about Hurd. I mean, a lot of those things implemented in kernel, sound like some userspace thing.

[skeevy420]

eBPF changes the way operating systems and infrastructure services are designed. It bridges the boundary between kernel and user space. It encourages and accelerates innovation and is a significant leap forward in open source technology for networking, security, application profiling/tracing and system observability use cases. eBPF enables users to even combine and apply logic across multiple subsystems which were traditionally completely independent.

I feel like I could replace eBPF up there with anything else and it would still work because that's just a bunch of buzzwords in the form of a paragraph.

I'm not saying this isn't cool or doesn't have potential.