/cyber/ - Cyberpunk & Science Fiction

>>50252

>Computer architecture proprietary, closed

>>49727

I am looking to build a non-botnet router soon. I am currently using a PfSense box running on an old Intel Xeon server that I am sure is vulnerable in some way or another. Is there any FOSS ARM box that either has multiple 1Gb/s interfaces, or supports 802.11Q for VLAN trunking? Also what OS is recommended for routing/firewalling? I don't really want to have to build IPTables and routing tables manually, the web interface of PfSense is very convenient.

>>50254

I'm a total noob, but i've seen banana pi version with multiple gigabit ports.

>>50263

The BPI-R1 or R2. I bought the R2 and i can't say i recomend it. They decided to ship before they figured out how to get all the hardware working. The offical image barely works and doesn't even support the onboard wifi. Now the best support comes from a small handul of install images created and maintained by some guy named greg on the bpi forums. If you're experinced in hacking around in linux and flying blind, it will work, but don't expect a smooth experince or even good network proformance without beating your head against a keyboard for several hours per-issue.

>>49749

>Let me save you lifetime of introspection. You don't do all the things you're sure you could do, because you're not the sort of person that does things, and at then of your life you're going to realize you lived the mediocre life of a porn addicted shut-in, who never actually did much of anything.

>Don't kill yourself, just keep telling yourself you're awesome, even though deep down you have to admit there's no proof.

This was very specific.

…do you need a hug?

>>50769

Not the person you're replying to, but it's not specific at all, it's the vast majority of people on the net.

If you want proof, next time people argue about some programming language, ask to see their git repo of projects they are doing in that language. I guarantee you won't hear back from them.

>>50777

That's not very fair, I'll evangelize C and Lua till the cows come home, but I'd never link to my repos (on the chans) because I have my name attached to most of my work.

Back on topic, I started investing in the stock market just so I could buy one of these. 1/7th the way there so far. I wish they'd come out with a more budget model. It would really help IBM get back into the chip game.

>>49727

There's too much marketing inside just do not be stupid, use calculator.

$ 5,149.00 Talos ™ II Secure Workstation included only $ 2,499.00 Basic Talos ™ II Bundle (Single CPU) so cost of case and Power Supply cost $ 1,700 ($ 7,284.00 with optional component so component cost include cpu is $ 2135 real like $ 1000-1200 with resonable alternative $ 700-800 )

$ 2,999.00 Basic Talos ™ II Bundle (Dual CPU) so only motherboard expected cost - $ 2199 or $ 2249 real which included x2 cooler for $ 50

$ 2,499.00 - Basic Talos ™ II Bundle (Single CPU) so only motherboard expected cost - $ 2099 or $ 2124 real which included one cooler for $ 25

$ 2,499.00 Talos ™ II Mainboard (Board Only) So CPU with headsink cost nothing.

$ 1,099.99 - Talos ™ II Lite Mainboard (Board Only)

$ 375.00 - IBM POWER9 CPU (4-Core)

$ 110.00 - 3U HSF Assembly for POWER9 CPUs

Most cheap option Talos ™ II Lite Mainboard (Board Only) + IBM POWER9 CPU (4-Core) $ 1,099.99 + $ 375.00 = 1474.99 or even two for "objective" cluster with similar perfomance 1474.99 x2 = $ 2949.98

>>55020

>amigafags will eat up the scraps of embedded Power cpus even after Talos II shipped

This never fails to make me laugh at them. Theyre worse than macfags; at least macs run on modern hardware meant for real computers.

>>55042

fucking 水 setup. Thats how you escape the matrix.

>>55043

I haven't bought shit except a rpi3 but, Talos II at 5k is outside any budget for a desktop computer I will ever, I just don't need 36 logical cores right now as 水 as that is. I want FreeRISC, but obviously Power is an entirely different beast. Entry level is twice the price of shitty amiga system. The amiga is something like ~2010 grade non-intel hardware - which I am using now, at half the price.

In order of price its: arm/freerisc soc(~20-40 USD), Amiga (1.5-2 kUSD), Talos II (5k). The mid range consumer grade hardware 500 - 1000 USD is totally missing, because this isn't a normal consumer market - but CPU diversity would be nice.

>>49727

>X86 Considered Harmful. What does a chummer use. What do you think ?

The question is almost done to death by now, if you have to use x86 for some reason use AMD, if you don't want that use a decent ARM based setup like >>49812

A POWER or RISC-V based system would be king but one is very expensive to get and the other has pretty much nothing aside from some dev boards.

>>55044

>水

I see you are a man of culture. My main point was that the single CPU Talos Board is more cost effective than the x5000: $1.5k vs$1.9k so about the same price after adding other components to the Talos' price. Librebootable x86 boards ought to be in that list of hardware between RISC SBCs and the Amiga because they can more than keep up with the Amiga at a midrange budget.

>>55051

>>>水

<pronounced schway in chinese i thought that shit is too 水 to pass up. Cityspeak.

Librebooted x86 is a good add to the list. I like the idea of CPU diversity and so I want to get away from x86, but If I ever buy x86 again, it will be a librebooted system.

>>55059

Can't you get some ARM sbcs with 4 or 8 gigs of ram now? I'm sure at least some are free of botnet. Pretty sure the FSF has a recommendation list on their site, and hnode seems like a good resource too.

>>55060

>ARM

TrustZone. This is what AMD based their IME-like "Secure Processor" on.

>>55062

Not every modern ARM implementation has it however. Do your research, and you'll likely be able to find a non-kiked ARM system with the processing capacity you need.

>>55063

I had no idea. Thanks Chummer.

>>55060

4 gigs is easy to find in the Rockchip-based board. If you got the money, this might be a good one:

https://www.theobroma-systems.com/rk3399-q7/tech-specs

It's a SoM that mounts onto this board:

https://www.theobroma-systems.com/rk3399-q7/dev-kit-haikou

Anyway they're an austrian company, so it's bound to cost more than the chinese stuff. Prices aren't listed, so dunno. A cheaper option would be this:

https://www.pine64.org/?page_id=61454

OpenBSD seems to support both, at least for the basic devices (no idea if stuff like audio works though):

https://www.openbsd.org/arm64.html

And just forget entirely about 3D GPU support, unless you get one of the boards with reverse-engineered driver, like indicated on this page:

https://www.fsf.org/resources/hw/single-board-computers

Personally I have completely abandonned the idea of needing 3D GPU at all, ever. That means I also don't play modern games, or even fancy emulators like MAME that now require GPU to run well.

>>55069

I'm not familiar with it, but sounds like it's an optional feature, and not something that's always on by default (like Intel ME). Sounds like it's up to the OS to make use of it, for virtualization, or maybe you could leverage it via a microkernel.

https://en.wikipedia.org/wiki/TrustZone#TrustZone_(for_Cortex-A_profile)

>>55070

My reading of >>55075 is basically the same. Its a technology that allows the creation of two "worlds" that segment memory with an extra address bit to enforce separation between them. A special instruction SMC is called to switch context, and the "secure world" can act as a software library, or a whole another operating system. Unlike TPM, there is no secured flash memory, and other resources like disks are not multiplexed securely between worlds - so there is no key storage. It seems like an "orthogonal" system of privileges to "ring levels". I don't actually see what I would want this in my computer - why do I need two orthogonal systems of privileges - at the same time, as long as I have control of what runs in the "secure world", i don't actually care - it could be used to build someschway stuff perhaps.

Its definitely not Intel ME by itself.

>trusting IBM

Hmm… no. Don't even give me that "quality" bullshit when you don't know what kind of experiments and weapons they use on mankind. They're the thinktanks of the elites and falling for this expensive trick will be your downfall.

>>49727

No microprocessor is safe. Everything is a grand design to tag the goyim as this planet only serves as a single purpose of ayy gold mine. Yeah and now that you know don't blame me if you see some grey ayys standing and staring at you out your doorsteps. Even I am risking my self by writing this and I know global TEMPEST snooping programs like HAARP they can literally see what I am currently typing and my fucking screen if not realtime then maybe a little later.

Enough is enough, we enjoy art and a lot of fantasy themes and contents made by none other than ourselves but the truth is this entire grand scheme is smoking a giant turd and these art are just clay figures of that turd. Our lives are a lie and what's worse is by the time we reach genetic "realization" of truth will be the time they reap the entire planet off its inhabitants. Trust Q maybe, until they release the declass. Don't know why it's taking long but it might just be another extinction for us, well if you manage to not get sucked in that gigantic soul machine then you should follow the stars and let your sign tell who you are, ask the stars and let them tell who you were, and let your immortality radiate into this giant turd. You know who you are, you can ask the stars who you were. Some say it's occult shit but in reality we really don't know but as long as it works its magic then that's it. Just do it.

Billions of cogs that need no repair,

Few hidden and mostly wanted.

Knowing be your ultimate despair,

Immortality be granted.

>>55076

Still doesn't sit well with me, but I'm overly paranoid.

>>55113

Architecturally, there are ONLY two worlds supported - a secure and non-secure world. If you have control of both - then you have control of the chip.

Although, how do you know the chip only supports two worlds ? ARM is nice enough to let you rich corporations across a diversity of nations inspect and modify the design of the chips - However, even if you were to see and validate the design, how can you tell what has actually been fabbed?

Not trusting your hardware vendor is a tricky road to go down. You can only really mitigate risk.

I think the chance the an ARM processor has a designed in backdoor is less than intel because more than just intel has seen and used the design - ARM is more open. However, there is no certainty. I don't think that arm processors without trustzone are less likely to contain a backdoor than arm processors with trustzone because, if your hardware vendor is going to maintain two designs, a secret internal backdoored, and a semi-public backdoor free design, they are going to do that trust zone or not.

>>55140

Fair point.

It's a shame there's no easy way to have truly secure hardware. I wish trust wasn't part of the equation at all.

>>55112

>Qoomers found cyber

There goes the neighborhood.

>>49727

>http://archive.is/PyXVk

> In System76's road to manufacturing their own laptops and desktops, the Linux-focused Denver-based company has their eyes on offering ARM-based products.

I would buy a System 76 arm laptop.

>>55203

HiFive requires proprietary firmware to work. Don't use this botnet.

Better buy this https://www.crowdsupply.com/eoma68/micro-desktop

>>55204

A computer small enough to maintain constant constant physical control of is very schway. Intel makes a similar device but obviously it's x86. You'd need to watch out for sniffers in the ports of the laptop/dock though.

>>55204

The thing is it doesn't matter. If there is no proprietary firmware that just means they burned it into mask rom. Any bad stuff that could be done with proprietary firmware can be done just as easily with proprietary hardware. At least parts of the sifive hardware design are partly open source, maybe eventually it could be fully open. In any case complaining about proprietary firmware while suggesting an allwinner board is pretty rich.

>>55290

There is a difference: malware can embed itself into firmware. The Kaspersky guys found such malware that hides in HDD controller firmware a few years ago. I don't have link so you'll have to search for it if you're curious.

>>55290

>At least parts of the sifive hardware design are partly open source, maybe eventually it could be fully open.

If the hardware was "open source" you could make a free software replacement fot the firmware, but you can't.

Sifive chips are nonfree hardware running nonfree firmware.

With EOMA68-A20, you have fully free firmware. Just the core is proprietary.

>>55294

That's a fair point, but it applies equally whether the firmware is free or proprietary.

>>55295

>If the hardware was "open source" you could make a free software replacement fot the firmware, but you can't.

Right, because it's not open source. The hope is that by supporting risc-v there could eventually be an entirely open source SoC. Meanwhile Allwinner is a regular GPL violator and dumps barely working code with the expectation that the open source community will fix it.

>Just the core is proprietary.

If by core you mean the entire SoC, which is basically the computer. The rest is just wiring, random peripherals and jelly bean parts.

>>55297

>If by core you mean the entire SoC, which is basically the computer. The rest is just wiring, random peripherals and jelly bean parts.

At least it can be run only with free software, excluding Mali GPU driver, but in EOMA68-A20 it just isn't included at all. Reverse engineering efforts are for Mali are in progress - Lima, Panfrost, etc.

>The hope is that by supporting risc-v there could eventually be an entirely open source SoC.

I'm shilling it all over the place, but here

https://www.crowdsupply.com/libre-risc-v/m-class

The same dude that is creating EOMA68, is working on Libre Risc-V processor and will eventually use it in EOMA68 compilant computer card.

>>55297

If it's trully open source then it includes tools and methods for reflashing the firmware. That way at least you don't have to throw away your hardware if it gets pwnd.

But I don't know any such HDD controllers, so if your shit gets pwnd (and by some miracle you actually become aware of it - not so easy with rootkits that dig in this deep), you can try to move the disk to another "identical" board from the same manufacturer (hard to tell what's identical since they may update/change the firmware and parts without notice as it's not considered "user serviceable"), or you can throw it out (this is the most likely case).

>>55297

>The hope is that by supporting risc-v there could eventually be an entirely open source SoC

That may happen, but never for the mid or high-end stuff. Only megcorps can afford dev+fab for that, and they won't touch a design/arch that's not proprietary. Just look at what happened to SPARC after OpenSPARC.

>>55300

HDD hacking is definitely something I need to look into more; if only to be able to dump and store "known good" hashes for the firmware

>>55443

>Only megcorps can afford dev+fab for that

Not him, but with Risc-V it is possible to create own core relatively cheap.

Also

>>55298

https://www.crowdsupply.com/libre-risc-v/m-class

There is one completely libre SoC being developed.

Just supporting Risc-V won't help, as using WWW which is a free standart too won't make all js scripts and all webpages free.

>>55464

>low-power, mobile-class

>suitable for tablet, netbook, and industrial embedded systems

For non-highend stuff, RISC-V all the way, but I'd want something properly powerful for my workstation/desktop

My plan is to eventually migrate all my stuff:

x86 -> Power

ARM -> RiscV

>>55465

>For non-highend stuff, RISC-V all the way, but I'd want something properly powerful for my workstation/desktop

Mobile-class processor is just the beginning, because it is easy and cheap to develop, the risk is lower and the demand is higher.

You could buy a few cards and connect them making a cluster though.

>>50232

Not likely, you already got boards going for reasonable prices. Bunch of companies are getting into this because it means not paying a pound of flesh to softbank for every arm chip you make. This means tons of companies will be making these.

There already a company that, no joke, makes custom RISCV system-on-chip, you can drag and drop parts online, check

<https://www.sifive.com/chip-designer

>>49727

>Secure workstation

<Doesn't list a TPM header in the specs

Ahhh, you almost had me you fucker you.

>>55537

https://wiki.raptorcs.com/w/images/e/e3/T2P9D01_users_guide_version_1_0.pdf

page 8 Mainboard specifications

> System Features: TPM Header: LPC / I2C