/sci/ - Soyence and Technology

the later

>>7675
esl trvthnvke

>>7675
That sounds like a waste of time if you're not hiding 'p or care about datamining.

>>7678
gnuschizo has confirmed that he has 'p on his computer.

>>7679
Does he live outside the us where their feds don't giver a crap?

>>7680
He is German where pedophilia is normal and people have sex in public places, on buses, parks and beaches.

>>7660
all freetards are cultists what do you mean?

>>7683
Is he a software purist for fun or does he have something to hide?

>>7684
>does he have something to hide?
yes, 'p

>>7685
'p is legal in his country though

>>7660
he is a free software cultist who thinks tails qubes whonix tor browser are all le evil backdoored or whatever even though not using tor browser makes your fingerprint stand out. like fuck I wish qubes could be FOSS but it needs to settle with blobs because it is a security based browser and the likelihood that the blobs are gonna getcha is much lower than the likelihood that you get exploited because you don't have them.

put it this way, Darknet markets exist and some of which have exited before without being caught, do you really think the feds aren't going to breach the market and arrest everyone when they have wasted trillions on the war on drugs?

>>7688
I am sick of the "blobs" nonsense. You will never be blob free. A "blob" is just a stupid word for firmware or microcode. ALL x86 CPUs are microcoded. Somehow, it's a violation of your "freedom" to not have a copy of the microassembly but it would not be a violation of your freedom if the same CPU was hardwired like RISC CPUs are. And what about hardware? All digital ICs are designed in VHDL or Verilog. Must IC designers make the VHDL freely available? And "privacy" is a cope. The rationalle of the FSF has nothing to do with privacy or backdoors but rather that it is somehow morally wrong to not give people source code because it just is okay?

>>7690
The gnuschizo will lose his mind when he finds out he's running nonfree Verilog

gtfo tofufag

>>7686
this
>>7685
i don't hide the fact that i have 'p though
im a freetard because i don't want to be fucking doxxed and shot by feds for doing illegal shit or just not acting like a good goy who eats the bugs and takes the vackseen
>>7688
i have the exact same fingerprint as torah brousah, and i don't have DRM bullshit in my browser
if you need a separate browser to use tor, you're mentally retarded.
literally, about:config, 127.0.0.1:9050 as SOCKS proxy, disable JS, and you're good.
>durr hurr i wish it could be hecking fuss
if you kicked out the blobs it would be, blobs instantly obliterate any freedom security or privacy you had whatsoever anyway.
qubes and tails are CIA nigger backdoors, so are all those other meme distros
lets take artix for example, commonly used by hecking privacy lovers who hate soystemdick
its super hecking free and private right?
well explain to me why the default guishit install phones home then
>BUT I NEED MUH BLOBS FOR MUH SHITTY HARDWARE
stop buying backdoored hardware, no hecking meme OS will save you if you have fucking hardware backdoors, retard
>proprietary = security
this is bullshit and literally made up by feds, you can't have privacy without security, and you can't have security without the freedom to know what the fuck you are actually running, and the freedom to actually control your computer.
>durr hurr how do all the hecking darknet kiddies get away?
because the feds dont fucking care if some nigger is selling weed for meme shitcoins without paying the kosher tax, they care if you're a terrorist, but that's no excuse for me to tolerate blobs even if im not a terrorist.
>>7691
im not though

>>7693
pedo cope. hope the feds shoot you.

>>7690
>You will never be blob free.
Only because cpu producers make their microcode closed source.
>A "blob" is just a stupid word for firmware or microcode.
wrong, it's proprietary firmware or microcode.
>that it is somehow morally wrong to not give people source code because it just is okay?
zzz you are retarded

>>7693
>i have the exact same fingerprint as torah brousah, and i don't have DRM bullshit in my browser
wrong.
>no hecking meme OS will save you if you have fucking hardware backdoors, retard
your special stash of cp is not worth multi million dollar cia hardware backdoor. kys. many DNM operators have used and likely continue to use Qubes.

>>7693
>this is bullshit and literally made up by feds
you're a retard, the blobs in qubes are patches to the CPU microcode, they are necessary to prevent known exploits against your CPU.

they would be better if they were open source, sure, but you are 10000x more likely to get exploited by malware vs cia backdoor in le hecking blobs.

>>7693
>because the feds dont fucking care if some nigger is selling weed for meme shitcoins without paying the kosher tax
wrong, factually false, backed up by nothing, unsubstantiated, and most of all, not peer reviewed by snopes

>>7706
All microcode is proprietary dumbass. If Intel just hardwired the CPU then you wouldn't be complaining. Why on hell would a CPU manufacturer make their CPU design open source? GNUtards in la-la land where R&D is completely free and there is no such thing as Club Goods. Do you know how closely Intel guards their IP? When a fab is decommissioned, it is shredded on-site and the different metals are sorted on-site. Hardware development is expensive. This is why companies don't provide you the microassembly. It's not because they want to put "backdoors" in your computer in the most convoluted way possible. Gnupedos are really stupid.

>>7710
i complain because ucode updates are like 50kb, while ucode etched onto the cpu is barely 1kb
you get a bunch of advanced new malware instead of old backdoors that might give you ring 0 access from running some hacky ring 3 shit or opcodes nobody even heard of
besides, why would you waste 20% of your CPU power just for muh mitigations to get more recent backdoors?

>>7711
The backdoors which still have never been demonstrated to exist. The microcode is not 1KB on the CPU. The microcode on the 486 was 12Kb. Microcode updates were implemented to fix issues like the Pentium F00F and FDIV bugs without requiring a recall of the product. If Intel wanted to hardcode a funny sequence of bytes to get Ring 0 from 3, they could have. Intel doesn't however because they aren't in the business of PR suicide.

>>7716
You don't know what you are talking about. This is a video made by Christopher Domas. This is not an exploit. This is a well documented feature on some VIA C3 CPUs known as AIS (which Domas calls the "deeply embedded instruction set" because he is a dishonest hack). So well documented that there was a proposal for NASM to add AIS instructions into the assembler in 2005 but was declined because NASM is an x86 assembler, not a AIS assembler. That "secret instruction" that the guy is showing is JMPAI, JuMP to Alternative Instruction. The bound instructions are not bound instructions at all bus AIS instructions. You can read the AIS documentation here: http://www.bitsavers.org/components/viaTechnologies/C3-ais-appnote.pdf
And before you say that it was hidden because it says confidential, all you needed to get the application note is ask VIA nicely and not be a retard (it effectively says that VIA is limiting the manual to non-retards in the manual). KolibriOS developers did the same thing to get the DM&P GPU datasheet.
And the function of ALTINST is documented in VIA publicly available datasheets as well:
http://datasheets.chipdb.org/VIA/Eden-ESP/Eden%20v1.4.pdf
The reason that this instruction set exists on VIA C3 is that VIA C3 CPUs are like P6 where they convert x86 opcodes to Centaur Technologies RISC instructions which are executed on the C3. On most C3 CPUs, ALTINST is disabled by default, however, on SOME Samuel 2 CPUs ALTINST was enabled by default. This is easy to fix though as you need to write the MSR that disables ALTINST.

Most of Domas's claims are false. For example he said that AIS: "isn't supposed to exist (wrong, VIA is very clear that it exists), doesn't have a name (wrong, AIS), and gives you root right away (wrong, only if you have it enabled)."

But wouldn't it still be a vulnerability if you didn't have it disabled by default? No. It wouldn't be. Windows NT disables ALTINST on startup in KiGetFeatureBits (which you can verify for yourself by looking at the WRK source code: https://github.com/mic101/windows/blob/master/WRK-v1.2/base/ntos/ke/i386/kernlini.c). So you are only vulnerable if you install an old version of Linux on a weird Taiwanese x86 CPU from the early 2000s found in mostly thin client applications that run Windows.

>>7716
Here is the full AIS instruction set documentation: http://www.bitsavers.org/components/viaTechnologies/C3-ais-reference.pdf

>>7733
in other word, its a public secret and they are trying to prevent people from figuring it out so script kiddies don't do fucky shit with x86 CPUs

>>7762
Basically yes because if everyone started to use AIS to push a tiny percent of performance out of a C3 CPU then VIA would have to start supporting AIS on all their future processors so that devs wouldn't complain that the feature, that is primarily for testing purposes and I bet Centaur thought it would be fun to put in the final design, was missing. Newer VIA CPUs do not have AIS at all because, from what I can tell, do not use a P6 style CISC-to-RISC translation. One off undocumented testing instructions that ended up being used for unintended purposes in the past are the LOADALL instructions (two different instructions, one for 286, one for 386) which allowed for funky descriptor cache modification and nonsense like real mode paging. All of these tricks stopped working when the 486 was released because the instructions were removed and anything that used those instructions and didn't check the model of the CPU and stepping would crash (mostly memory managers). Centaur (and by extension, VIA) advertising AIS would have caused them to become pigeonholed into making the same microarchitecture.

One off design decisions end up having a big impact later on. For example, all x86 CPUs emulate an AND gate which is controlled by a i8042 keyboard controller connected to the 21st address line because theoretically, the user could be running 16-bit real mode software that takes advantage of the last segment of memory wrapping around. Picrel

>>7770
god x86 backwards compatibility is so wasteful I hope x86S becomes reality

>>7774
x86s = evil. segmentation is a human right.

>>7775
what do you mean by this

>>7660
he has to care about opsec because he's storing 25 petabytes of 'p

>>7778
Retvrn white man