- cross-posted to:
- security@lemmy.ml
- cross-posted to:
- security@lemmy.ml
If GrapheneOS stops supporting future Pixels, and Motorola is only targeting the flagship market for their compatible phones, we’re looking at truly dark days ahead.
If GrapheneOS stops supporting future Pixels, and Motorola is only targeting the flagship market for their compatible phones, we’re looking at truly dark days ahead.
Not complaining about Motorola, just about flagship phones in general. I assume (incorrectly, maybe?) that the vast majority of people use their phones for light browsing, social media, etc. and don’t need flagship features. Having to pay a flagship premium would suck… especially when all I really want is my Pixel 4a5g back lol.
Graphene requires flagship hardware.
I don’t mean like graphene as it stands now requires flagship hardware, I mean that graphene as a concept requires what is priced and treated as flagship hardware.
Compare the processors in flagship phones and those in gimmies by looking at their supported features and you’ll see a significant difference not just in cores and cache and onboard memory but in actual documentation and support for the kinds of hardware features (like memory tagging as referenced in this post) that graphene relies on.
Flagship phones don’t integrate and support these features because the manufacturers would like to give you what you’re paying for, it’s because meeting the needs of organizations that require them for their device fleet is profitable. If that component of phone design and engineering didn’t exist then everything would be Samsung whose device line can be fully differentiated by their screen sizes, camera features, screen resolutions and battery lives.
At the gimmie end of the spectrum it is already like that even for iphones although not as much.
Google was using the pixel line as a reference implementation of the Android platform, with their goal of providing device manufacturers something to measure their own work against and pay attention to to the function of so their android os wouldn’t become a flash in the pan and they could extend onto mobile use as a company. Android use grew significantly and they had to shift to using the pixel line as a trellis for a vigorous garden plant.
Bizarrely the pixel line was a money loser or a breakeven proposal for Google for several years. When you think about it that makes sense: they’re not playing the same game as the other Android manufacturers (make the cheapest thing, sell the cheapest thing!) so of course they’re not going to be looking good by the metric you use to measure that. The new Android phones represent a break with that plan and an attempt to move to competing directly with other major manufacturers so now their phones aren’t any good for graphene just like the Samsungs of the world.
Now a person might say “why can’t the damn phone companies just give us a gimmie phone with advanced security features we want?!” Well such a thing is possible but it would be a money loser from the drawing board and wouldn’t fit easily into the rubric of phone sales and tie up a shit ton of resources and time. So even if someone at a hardware company decided to do that their boss would shitcan the project asap.
That same person might say “well why can’t graphene support a gimmie phone?!” And that a great question too but it would mean diving into hardware that doesn’t have the capacity to meet the bar graphene has set for itself in terms of security. Hopefully the person who started that project would stop themselves when they recognize the situation.
We’re talking about security features here, not processing speed. There’s no excuse for even the cheapest SoCs not to support them. Withholding them from the bottom end of the lineup is pure product differentiation (a.k.a. profiteering, a.k.a. spite and contempt).
It’s really easy to come to that conclusion.
Realistically speaking a team is trying to hit a set of requirements under a set of constraints when they design a soc (or any cpu if you wanna dive into the cisc world and see the wild variation in security featureset across even the same generation of chips). That means using existing designs with a die shrink or something is often the right choice when you’re shooting for the low cost portion of the market. The technology referenced here, enhanced memory tagging extensions, is a hardware implementation of tagging that requires a specific revision of the arm design and it’s a relatively new change (mte, the precursor to emte, is flawed and not used).
So it’s not by dint of wanting to fuck you over that the designers of socs would make one without emte.
I’m writing this on a Pixel 8 running GrapheneOS, which I’m pretty sure is a “recommended” device that supports all the security features and is also multiple generations old.
If new lower-end devices use die-shrink versions of previously flagship SoCs, as you say, then I simply do not buy the notion that there’s any good reason not to put a die-shrink version of this SoC in a new lower-end phone.
I don’t have the datasheet in front of me but I’m almost 100% sure your tensor doesn’t have emte, which is the type of mte the graphene people are requiring for new devices.
From https://grapheneos.org/faq#recommended-devices :
Why would this idea be a money loser?
Because it costs more than it would make!
Here’s how socs are: there’s some spec and some company develops a soc platform to address a portion of the spec. They come up with high priced top of the line socs to meet the full portion of the spec they’re trying to address and bottom of the barrel socs to address some much smaller portion of the spec they’re trying to address. They have some socs in between that fulfill different use cases.
Now look at phones, a flagship phone can either be a value device (best screen, best camera, fastest processors with the most cores) or it can be a Swiss Army knife (completely waterproof, supports every protocol, has some now esoteric port). You might imagine a different kind of flagship phone but a flagship phone can’t be what you imagine because a flagship phone is by definition the phone people will pay the most money for.
Of course people will only pay the most money for the phone that’s either the best thing or the most thing.
Gimmie phones are the opposite end of that spectrum. They are intended to meet the absolute minimum requirements to be considered a cell phone that the provider can give to someone for free (the cost isn’t up front but is rolled into the contract, in this way the service provider is under intense pressure to sell the cheapest gimmie phones to maximize their margin and rejects those that are either marginally more expensive or not desirable by their customers) if that person signs a contract with them.
Under those design constraints the featureful soc is verboten. You gotta go with the small very limited soc to meet the cost constraint.
A very small very limited soc won’t have a to the letter implementation (or any implementation) of the security features required to run graphene. Things that are implemented but not required for the mast common use cases are not documented or troubleshot.
Any cell phone will have at least three socs in it, all working like that.
You can imagine a gimmie phone that supports graphene but it won’t happen because of the systemic forces at play.
They’re not targeting flagship phones, it’s that the SoC the flagship phone comes with is the only SoC that meets the security requirements. As that SoC ends up in the next gen phones, cheaper options will be available.
https://www.techtimes.com/articles/325679/20260826/grapheneos-expands-motorola-better-snapdragon-radio-isolation-google-pixel.htm
Cheaper options will be available in the future. Let’s just be happy hardware will be available period.
Potential loss of profit from data sucking. Corporate fleets with something to lose will pay more, but in a cheap thing for the masses? Sell their data / habits/ souls…
True. The flagship thing makes it a bit out of reach for me.
The flagship thing just makes it daft for me, a phone is just a phone for some of us. I have a pc.