[LinAGKar]

So basically to summarize, Google embargoes security patches for four months so OEMs can push out updates more slowly. And if those patches were immediately added to an open source project like GrapheneOS, attackers would gain info on the vulnerabilities before OEMs provide updates (the GrapheneOS project can see the patches, but they can't ship them). But a lot of patches end up being leaked anyway, so the delay ends up being pointless.

[lima]

The stupidest part is that, according to the thread, OEMs are allowed to provide binary only patches before the embargo ends, making the whole thing nonsensical since it's trivial to figure out the vulnerabilities from the binaries.

Fun fact: Google actually owns the most commonly used tool, BinDiff ;)

[nroets]

Unless the OEMs bundle numerous changes with the security patch(es).

(I'm not saying it happens. I just theorise how the policy could have been envisaged)

[numpad0]

In the good old days, there were exploits patched years prior by some OEMs that were never upstreamed even to Google. New rooting apps come out and... just doesn't work. I don't know if that still happens, though.

[groggler]

Not really.. numerous changes are still not a total redesign of whichever subsystem was affected so it's pretty obvious where some small security relevant changes are. A stupid embargo was always enough to ruin security by code analysis for white hats but never enough to stop attacks by code analysis for black hats.

[lima]

Yep, people have been reversing huge Windows patches for decades and it never stopped anyone from figuring out what the fixes were.

[tester89]

How does this work legally? If Android AOSP is open-source, once one OEM updates, surely the owner gets the legal right to request sources. IIRC the maximum delay is 30 days.

[bri3d]

Almost all of AOSP is under the Apache or BSD licenses, not the GPL. Very few GPL components remain (the kernel being the large and obvious one).

So, yes, making a GPL request will work for the very few components still under GPL, if a vendor releases a binary patch. But for most things outside of the kernel, patch diffing comes back into play, just like on every closed-source OS.

[dijit]

weird tangential question then: when does GPL stop being infectious?

I would understand in a modular system like an operating system: one can argue that the kernel is a single component.

But if you're buying an appliance, the OS is effectively one single unit: all linked together.

Why does a binary executable and a binary image seem to operate differently in this space - both are inscrutable?

[rollcat]

The FSF has always been pretty clear on this: you use a linker (static or dynamic) = it applies; you don't = it doesn't. They even wrote LGPL with this distinction in mind, and introduced exceptions to yacc (bison) to accommodate non-free software.

In case of binary releases, you can request the sources of the relevant subcomponent (e.g. the kernel). The component boundaries are pretty clear wrt Linux: Torvalds has made it quite clear early on, that the kernel's GPL2 does not apply to anything in the user space.

Here also, the important distinction between GPL 2 & 3: with GPL3, it would be a breach of the license to ship code on a device that does not allow the end user to update that code. Which has effectively pushed everyone away from GPL3-licensed software.

IMHO the move to GPL3 has likely caused more harm than good to the FOSS ecosystem; in some alternative universe, GPL3 never happened, most of Android's userspace is GPL2, and we get the source for everything. In both universes we still don't get to deploy changes to devices we own, so IMHO the GPL3 won us nothing.

[cyphar]

The FSF considers linking to be a definite example of derived works in general, but I don't believe they consider lack of linking to prove that something isn't a derived work.

The goal of the GPL is to flip draconian copyright maximalism on its head, and copyright laws don't talk about linkers so that can't be the deciding factor. Not to mention that it would be trivial to work around linking by creating a stub and calling the GPL code as a subprogram (in kernel contexts a spiritually similar setup is called the "GPL condom" and my impression is that most lawyers not employed by NVIDIA consider this to not be a get-out-of-jail-free card).

[LegionMammal978]

> (in kernel contexts a spiritually similar setup is called the "GPL condom" and my impression is that most lawyers not employed by NVIDIA consider this to not be a get-out-of-jail-free card).

The whole thing with Linux's conception is that it's predicated on any and all unlicensed usage of GPL-only interfaces being copyright infringement of other usage in the kernel source. This is an extremely broad claim to make in general (especially in light of Google v. Oracle), and the 'GPL condom' approach is just to further ensure that the unlicensed side is textually unrelated to the kernel. When there's no infringement, the copyright holders can't do a single thing, except to technologically make it harder on you.

Meanwhile, the whole GPL idea of linking vs. statically embedding is only applicable when you're shipping someone else's GPL-licensed code alongside your non-licensed code, in which case you're bound by its terms. If you're not shipping someone else's code, then there's plenty of ways to force a particular build, etc., in the manner that the GPL is trying to prevent. Heaven knows I've likely violated the spirit of the GPL before just through Hyrum's law.

[raron]

That is explicitly mentioned in GPL FAQ:

https://www.gnu.org/licenses/gpl-faq.html#MereAggregation

[pjmlp]

Note that all FOSS OSes for embedded systems none of them are GPL based, Linux kernel is the only remaining GPL project amongst the alternatives.

Eventually everything comes to past, us in this realm, and the legacy left behind, eventually other projects will take Linux place, it can even be an heavily forked Linux.

[cyphar]

The short answer is copyright law and jurisprudence. The whole purpose of copyleft is to flip draconian copyright regimes over and make them protect users instead, so the GPL generally has the most maximalist stance allowed by copyright law. If copyright law would say that combining or extending software in a particular way is not fair use then the GPL generally would render the combination GPL.

In practice, GPLv2 would not be viral in the way you describe unless you can show that all of Android is a derived work of Linux (not true). GPLv3 would require users be avle to replace components under said license which has an impact on how such an appliance need to work (though the GPLv2 does also have somewhat related text about "the scripts which control installation") but wouldn't expand the scope of code under the license, just the terms.

[immibis]

Sadly, the layperson's and lay developer's interpretation of the GPL has been watered down over the years, and the GPL wasn't maximalist enough to begin with - see AGPL, SSPL for extensions created by people who saw new kinds of linking that didn't appear to be covered by the GPL. Of course big corporations preferably use these new kinds of linking that aren't covered, which is why new licenses are necessary.

[nottorp]

> the OS is effectively one single unit: all linked together

If your appliance runs linux it has separate components just like desktop linux.

You want to do as little as possible in kernel space, and depending on the appliance there isn't even any need for it.

So, like desktop linux, you can have closed source binaries on top of the kernel.

[dijit]

I just don’t see the distinction as clearly when it’s a single binary that cannot be decoupled or introspected.

Why is it if I build a static binary with GPL code and distribute it I must open source my changes; but if you do the same as a whole OS it’s not necessary.

Feels like it should all be fine or none of it is fine somehow.

[nottorp]

If you go that way, every bash script you've ever written should also be under the GPL.

And R code, at a quick glance.

[neobrain]

There's a concept of "separate works", see for example https://www.gnu.org/licenses/gpl-faq.html#GPLCompatInstaller .

Tangentially, I assumed that the GPL must have some built-in exception for running non-GPL userspace programs on top of a GPLed kernel (similar to the System Library exception). However, it seems like it doesn't, since the Linux kernel has its own exception to allow this: https://spdx.org/licenses/Linux-syscall-note.html.

[cyphar]

Note that the Linux syscall exemption is actually not the license of the entirety of Linux, because most code contributed to Linux is under the standard GPLv2. It's just a red herring -- there is no need for such an exemption because the generally held view is that such programs are not derived works of Linux (from a copyright law perspective) in the first place.

[lmm]

> when does GPL stop being infectious?

Either a) when the license has an explicit exemption (such as at glibc or the kernel's userspace interfaces) or b) when something ceases to be a "derivative work" in copyright terms (which is ultimately a legal question for lawyers).

[surajrmal]

Or existing on the other end of ipc.

[lmm]

No. The FSF takes the position that a program that is deeply entangled with another is a derivative work and can be subject to the GPL, even if that deep entanglement takes place over IPC.

[immibis]

Have you ever tried requesting the source code for your phone?

They'll either ignore you, or give you something that is obviously not the source code (e.g. huge missing sections; often they'll only produce kernel code and not even a way to compile it). Law be damned. They don't follow it and nobody is forcing them to

[Hizonner]

Fuck, and I cannot emphasize this enough, the OEMs.

I am so sick of security being compromised so stupid, lazy people don't have to do their jobs efficiently. Not like this is even unusual.

[microtonal]

I don't think it is laziness per se. It's a combination of having far too many models (just look at Samsung's line-up, more than ten models per year if we don't count all the F and W variants), using many different SoCs from different vendors (just taking Samsung again as an example, using Qualcomm Snapdragon, Samsung Exynos, Mediatek Helio, Mediatek Dimensity, sometimes even a different chipset for the same phone model per region), each model supported for multiple years now on a monthly or quarterly update schedule (Samsung: recent A5x, Sxx, Sxx FE, Z Flip x, Z Flip 7 FE, Z fold x, Xcover x, etc. are on a monthly schedule). This across a multitude of kernel versions, AOSP versions (for older phones), OneUI versions (for phones that haven't been updated yet to the latest OneUI).

The must have literally over tens of different models to roll out security updates for, with many different SoCs and software versions to target.

And compared to other Android vendors, Samsung is actually pretty fast with updates.

It's true that other manufacturers have smaller line-ups, but they also tend to be smaller companies.

Compare that with Apple: every yearly phone uses the same SoC, only with variations in simpler things like CPU/GPU core counts.

[WorldPeas]

To me this is the ultimate failing of ARM as an ISA, the fact that you even need to consider "targeting" allows a deficient ISA like x86 to still stand head and shoulders above it in terms of OEM support (though perhaps not security)

[rollcat]

It has nothing to do with the ISA and everything to do with the system architecture. Look up PC-98.

Also: PC being a "standard" is a lie; ACPI is a horror.

[Hizonner]

> I don't think it is laziness per se

You forgot the "stupid" part.

> It's a combination of having far too many models (just look at Samsung's line-up, more than ten models per year if we don't count all the F and W variants), using many different SoCs from different vendors > [...] > This across a multitude of kernel versions, AOSP versions (for older phones), OneUI versions (for phones that haven't been updated yet to the latest OneUI).

Those are choices. If you want to do that, you need a process that can support it.

I suppose it could be that they just don't care and are deliberately screwing their users, but never attribute to malice that which can be explained by incompetence and all that.

[yathaid]

>> Those are choices. If you want to do that, you need a process that can support it.

__need__ is doing a lot of work here. There is no forcing function to get OEMs to do this ASAP: 1) the market doesn't really care that much 2) there are no regulations around this (and even if they were, can you immediately recall a tech exec going to jail for breaking the law ... )

[palata]

> the market doesn't really care that much

This. Pixels are not more expensive than flagship Samsungs. If people cared and bought Pixels because they get the security updates, then Samsung (and the others) would follow. But people don't care, so the OEMs don't do it.

[microtonal]

It's kinda weird to single out Samsung here, because they are pretty good with security updates and they explicitly talk about long security periods in their marketing. They are not as fast as Pixel, but somewhere mid-range and up (A5x) get monthly updates and they are usually 1-4 weeks behind Google.

It's the other vendors that are the issue. Even Fairphone is behind a lot (and they only release one model at a time).

[microtonal]

Those are choices. If you want to do that, you need a process that can support it.

I suppose it could be that they just don't care and are deliberately screwing their users, but never attribute to malice that which can be explained by incompetence and all that.

I think for a long time Android users did not really care. Until a few years, Android security support was abysmal with many vendors only doing 1-2 years of updates. Users bought the phones and didn't care, so I guess it was a smart business move to not care.

This changed in recent years due to a mixture of the (then) upcoming EU requirement for supporting devices multiple years with security updates, Apple being able to tout this as an advantage, causing Google and Samsung to enter into a competition to promise the largest number of years of security support, etc.

[Zigurd]

Welcome to Android. It started out a bit undercooked and Google relied on OEMs to make finished polished products. Then the reality that OEMs suck at software hit them in the face. They spent years acquiring more control of their platform while trying not to piss off Samsung.

[tracker1]

Pretty much this... and even then, they still suck hard. Apple was right to start off with as much control over their platform as they did. The only reason I never went with iPhone is it started as an AT&T exclusive, and you couldn't pay me enough to be their customer ever again.

[egorfine]

> the delay ends up being pointless

Why though? It is pointless from the engineering and security standpoints, but for Google this may serve their goals very well.