Freedom and security issues on x86 platforms • Hacker News Board

Re: Freedom and security issues on x86 platforms

#211

Post by Ericson2314 » Mon, Apr 04, 2016, 10:00 PM UTC

nickpsecurity wrote: post #151

Ericson2314 wrote: post #145

Earlier quoted context omitted.

Is there are repository for this? Or do they still do the cathedral model?

I'm pretty sure it's the catheral model. I'm not even sure that they're doing open-source with Leon4 onward as pretty much nothing happened with GPL'd Leon3 and GRLIB. Comp Sci people and companies doing rad-hard, space apps are still getting and building on it. Best way to deal with them is to straight-up license their tech for a Pi- or router-style board. Then fab, assemble, and sell that joker. That gets the ecosy…

I've done school-size CPUs with http://www.clash-lang.org/ --- it would be fun to convert someone's "real-world" design into Haskell with it. 'Twould really show off that order of magnitude code size reduction :).

I suppose cathedral vs bazaar doesn't affect that at all, but experience has ingrained in me "source tarball ==> won't easily build" biases.

Re: Freedom and security issues on x86 platforms

#212

Post by DominikD » Mon, Apr 04, 2016, 10:03 PM UTC

kristoffer wrote: post #195

nickpsecurity wrote: post #144

Earlier quoted context omitted.

Because the chip-makers and chip buyers are using ARM and MIPS in power-efficient SOC's. They could do the same with SPARC. They just didn't for whatever reasons. Far as power efficiency, kristoffer might be able to chime in as it's not in the data sheets for Gaisler. That's suspicious: either the numbers are bad or they leave it off given its meant for customization. Anyway, the Leon4... http://www.gaisler.com/index…

Power consumption is of course very much dependent on the chosen fabrication process and SoC configuration. A LEON3/4 core is comparable to something like a ARM Cortex-M7 and it is not the ISA (when comparing ARMv7 vs SPARCv8) but the implementation that will affect power most. LEON is quite small and power efficient.

Node selected for fabrication does not equalize power consumption. You could go to the same node with other, inherently more power efficient architecture and gain even more oomph per watt. Power efficiency stems from the architecture itself; manufacturing process is a red herring (and a costly one). What you're saying is pretty much like "seasoned bodybuilder would kick white-belt karate practitioner's ass, so it's clear that bodybuilding is better than karate."

Compare things that are alike. If you take a manufacturer (say: TSMC), pick its node (say: 16nm FF+) and you decide on a package (physical manifestation of RTL primitives in the silicon) you get better performance per watt on one architecture over some other. ARM and MIPS are inherently very power efficient. You can't just take SPARC and make it more power efficient than these two. It doesn't work like that.

It's also not true that ISA doesn't matter. ISA impacts bandwidth requirements heavily. This in turn impacts latency and latency hiding, cache requirements and many other things. In fact data transfer is typically as costly as (if not more expensive than) computation. Getting data to all the right places on the schedule eats power like crazy. This is exactly why ARM has Thumb. It's not like internally core does different things than it would do with wide ISA. It's just that stuff's more densely packed, which helps tremendously.

Which brings me to my last point. There's an open architecture that's quite nice. It's SuperH (or SH2 in its open source form), which in turn is what ARM's Thumb is based on. It's not perfect, but it's pretty solid. Omitting it in the OP makes me think author isn't very thorough with his research. But everything has to start somewhere. ;)

Re: Freedom and security issues on x86 platforms

#213

Post by kabdib » Mon, Apr 04, 2016, 10:03 PM UTC

duskwuff wrote: post #201

kabdib wrote: post #57

Earlier quoted context omitted.

The boot ROMs on these things are proprietary, though. The vendors I've talked with have been extremely coy about what's in them (got an overview from Marvell once, spent a day looking at code on a projector screen and getting a walkthrough. They could have hidden much).

The boot ROM typically gets out of your way pretty quickly, though. At worst it means you have to deal with some firmware-signing nonsense before chaining into a Linux kernel (or U-Boot); it isn't active in a running system.

I should add that the code we saw may not have been the code that was actually run. It had reset vectors and whatnot, but that's no guarantee there were no hidden ROMs that ran code before the code we reviewed got run. And it's certainly no guarantee there are no hidden hardware-level state machines that unlock . . . things. Things like ignoring X bits in pages, or being able to do some low-bandwidth computation with code embedded at the stenographic level.

Re: Freedom and security issues on x86 platforms

#214

Post by petra » Mon, Apr 04, 2016, 10:07 PM UTC

nickpsecurity wrote: post #133

petra wrote: post #129

Earlier quoted context omitted.

Nick, The larger context of all this issue is defense. So on one side there are the five eyes governments wanting it this way. On the other side(and probably very interested in 100% security), you might have various countries supporting terrorist organizations, terrorist organizations, crime syndicates, russia, china, etc. Doesn't this context hints to us that 100% security would be much harder than creating some des…

I've addressed a good chunk of what it takes for system and hardware security in at least two places: http://pastebin.com/y3PufJ0V https://news.ycombinator.com/item?id=10468624 There won't be 100% security because underlying physics fights you and our field is too new. Best we can hope for is making attacks hard and physical . There's great work in secure HW/SW architectures that should knock out about all SW stuff w…

Ok. Maybe that may work. But what about legal risks? extra-legal risks(like vanishing in the dead of night) ? soft risks - how would the wife of someone who is just the customer will respond when guys in black suits will come to her home ?

Or if you're method will work so well, are you sure TSMC/Samsung will even accept you as a customer ?

Because it doesn't seem like something that could scale without the legal/political side and that's really much harder than the tech(which is hard, no doubt).

Re: Freedom and security issues on x86 platforms

#215

Post by Sanddancer » Mon, Apr 04, 2016, 10:10 PM UTC

makomk wrote: post #100

mmebane wrote: post #65

Earlier quoted context omitted.

I've been using a Raspberry Pi 3 for the past week, and have been pleasantly surprised by the performance. It's no speed demon, to be sure, but it's good enough for all my basic tasks. I wish there were a general "open computing" branch of the Raspberry Pi Foundation that would produce a $50-$100 "pro" version with more RAM and faster bus+peripherals.

The Raspberry Pi still relies on a closed-source blob running on a CPU core whose instruction set isn't publicly documented to even boot, but I suppose at least it's possible to reverse-engineer that unlike Intel ME.

Broadcom released a considerable amount regarding the Videocore IV a couple years ago. Nobody's finished writing an RTOS for it quite yet, but the ISA is now documented.

https://www.broadcom.com/docs/support/videocore/VideoCoreIV-...

Re: Freedom and security issues on x86 platforms

#216

Post by em3rgent0rdr » Mon, Apr 04, 2016, 10:15 PM UTC

5ilv3r wrote: post #80

em3rgent0rdr wrote: post #63

Earlier quoted context omitted.

I'm using ASUS C201 which uses ARM. It was a Chromebook, now running libreboot and parabola linux.

Can you share documentation? I have an upstream kernel running on my peach pit, but have not replaced uboot yet.

https://libreboot.org/docs/install/c201.html

Re: Freedom and security issues on x86 platforms

#217

Post by em3rgent0rdr » Mon, Apr 04, 2016, 10:21 PM UTC

andrey_utkin wrote: post #146

em3rgent0rdr wrote: post #63

Earlier quoted context omitted.

I'm using ASUS C201 which uses ARM. It was a Chromebook, now running libreboot and parabola linux.

How is it regarding GPU driver (or what Mali is)? Are you running binary mali driver or FOSS lima?

I don't really do GPU-intensive stuff on it. It renders webpages & youtube fine, which is probably the most graphical intensive stuff. I'm currently using it mostly for porting MuseScore to arm, mainly using qtcreator for development.

The external hdmi doesn't really work properly...I remember not being able to run external monitor in full screen, or at the same time as the lcd screen.

I just installed and ran glxgears. It gets 250 FPS and a default window size. But I do get an command line error libGL error: unable to load driver: rockchip_dri.so, so I don't think I'm getting GPU.

Re: Freedom and security issues on x86 platforms

#218

Post by pc2g4d » Mon, Apr 04, 2016, 10:39 PM UTC

qubex wrote: post #75

xenadu02 wrote: post #31

Earlier quoted context omitted.

Sounds like rich, fertile ground for the NSA, KGB, and other state agencies. They could be deploying such code right now and I'm not sure we would know it.

The KGB no longer exists. It was superseded by the FSB in 1995.

Not entirely---it's still around in Belarus. And Transnistria and South Ossetia, too, though those may be more of imitators than actual remnants of the original.

Re: Freedom and security issues on x86 platforms

#219

Post by rsync » Mon, Apr 04, 2016, 10:44 PM UTC

nickpsecurity wrote: post #167

rsync wrote: post #155

Earlier quoted context omitted.

"Oracle's T1 and T2 cores are open-source to study." If you had to pick a Oracle (Sun?) T2 based system to purchase off of ebay, with the interest in using it as a "more free, more open" system, what would you buy ? What OS would you run on it ?

I wouldn't. I'd use Gaisler's immediately because it's fully open and already FPGA qualified. I'd then buy a good FPGA board. Then I'd run it on there. It would probably run like a multi-core version of my old Pentium II. Yet, I programmed, hacked, gamed, and so on with it. Later, I'd put it on an eASIC Nextreme or actual ASIC if money came in for better performance, power, and unit pricing.

"I'd use Gaisler's immediately because it's fully open and already FPGA qualified. I'd then buy a good FPGA board. Then I'd run it on there. It would probably run like a multi-core version of my old Pentium II."

Sorry, let me clarify ...

Pretend you have three kids. But at the same time you'd like to tinker with a fully open system from loader on up.

Is there an old sun sparc that would make rms happy that I could buy on ebay ?

Re: Freedom and security issues on x86 platforms

#220

Post by 5ilv3r » Mon, Apr 04, 2016, 10:55 PM UTC

occamrazor wrote: post #159

wtallis wrote: post #127

Earlier quoted context omitted.

Aside from the stale Mac Pro, which current Macs do you think are indicative of Apple giving up on performance?

The new macbook 12" is thinner and lighter than the older MBA, but without significant performance improvements.

Dude it has a 5 watt processor and a mobo the size of a raspi. I don't think their featherweight offering is appropriate to enter in the performance fight.