Page 1 of 4

Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 10:27 am
by zahal
INTEL needs to sell new processors. What do they do? Make your current processor unsafe and 30% slower!

https://www.zerohedge.com/news/2018-01- ... amd-surges

"Fundamental design flaw". Class action suit lawyers must be salivating over this one!

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 10:58 am
by commodorejohn
This was down to the IME or I'll eat my hat.

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 11:35 am
by jan-jaap
commodorejohn wrote:This was down to the IME or I'll eat my hat.

Start eating, it has something to do with speculative execution from userspace resulting in leaking information from more privileged protection rings and/or defeating KASLR : https://www.theregister.co.uk/2018/01/0 ... sign_flaw/

But it's got to be good. The anger can almost be tasted in this comment from LKML:

Code: Select all

2) Namespace

   Several people including Linus requested to change the KAISER name.
   We came up with a list of technically correct acronyms:

     User Address Space Separation, prefix uass_
     Forcefully Unmap Complete Kernel With Interrupt Trampolines, prefix fuckwit_

   but we are politically correct people so we settled for

    Kernel Page Table Isolation, prefix kpti_

   Linus, your call :)

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 12:04 pm
by Trippynet
zahal wrote:INTEL needs to sell new processors. What do they do? Make your current processor unsafe and 30% slower!


Not mine, replaced my Core i5 system with an AMD Ryzen last month 8-)

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 12:07 pm
by Raion-Fox
MIPS is unaffected :p

I do not buy AMD because I cannot take the performance hit (Benchmarks of Ryzens show abysmal single thread performance compared to Haswell Xeons) and because I do not wish to support their poor quality chips.

That all being said, I'm glad I'm buying a Talos. Because AMD still has issues including the Security Processor which cannot be disabled (the BIOS switch is just a dummy switch similar to the AMT switch in an Intel BIOS) as it is integral to boot.

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 1:13 pm
by commodorejohn
jan-jaap wrote:
commodorejohn wrote:This was down to the IME or I'll eat my hat.

Start eating, it has something to do with speculative execution from userspace resulting in leaking information from more privileged protection rings and/or defeating KASLR : https://www.theregister.co.uk/2018/01/0 ... sign_flaw/

Good thing I don't own a hat.

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 1:49 pm
by uunix
commodorejohn.. number of posts.. 666.. number of the beast... the same!! You do the maths..

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 2:10 pm
by commodorejohn
667 now!

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 3:35 pm
by vishnu
So wait, every Intel x86_64 CPU for the last 10 years has this flaw, even the newest ones? I was just about to buy an i5-8600K, is Intel going to offer fixed CPUs that fit the same socket? They don't appear to be too forthcoming with whatever they've got in mind from here on out...

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 4:16 pm
by Raion-Fox
I would hold off Vishnu until we get more info.

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 4:18 pm
by bifo
vishnu wrote:So wait, every Intel x86_64 CPU for the last 10 years has this flaw, even the newest ones? I was just about to buy an i5-8600K, is Intel going to offer fixed CPUs that fit the same socket? They don't appear to be too forthcoming with whatever they've got in mind from here on out...


Yes, and :lol: if you think Intel is going to try to replace 10 years worth of chips when they can just patch the OSes and let the owner take the performance hit. They've accused AMD and ARM of suffering from the same problem, something which AMD has flatly denied and ARM is being cagey on (which seems odd, ARM being a totally separate family line, but then Intel is a big customer...).

As far as I know, it's a flaw with the processor itself, while the IME is a chipset thing and wasn't put into most consumer-grade chips. The danger there was that it was in every server and its a black box, and AMD still has their own version of it with the same basic issue: they won't tell anyone anything about it. Last I heard on that was that the operating systems that run on both should be theoretically deletable because it was put there due to government security contract requirements and they, presumably, have something they've made in-house. One hopes.

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 4:47 pm
by Raion-Fox
bifo wrote:Yes, and :lol: if you think Intel is going to try to replace 10 years worth of chips when they can just patch the OSes and let the owner take the performance hit. They've accused AMD and ARM of suffering from the same problem, something which AMD has flatly denied and ARM is being cagey on (which seems odd, ARM being a totally separate family line, but then Intel is a big customer...).


Because AMD's stock rose 6% after it became public.

bifo wrote:As far as I know, it's a flaw with the processor itself, while the IME is a chipset thing and wasn't put into most consumer-grade chips. The danger there was that it was in every server and its a black box, and AMD still has their own version of it with the same basic issue: they won't tell anyone anything about it. Last I heard on that was that the operating systems that run on both should be theoretically deletable because it was put there due to government security contract requirements and they, presumably, have something they've made in-house. One hopes.


The Intel ME has been a part of all chips dating back to the Nehalem uarch Bifo. This issue we're seeing affects the branch prediction part of the CPU which has been a feature in Intel for more than a decade.

On another note it baffles me why people trust AMD so implicitly because they're as greedy and shady as Intel

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Wed Jan 03, 2018 8:16 pm
by robespierre
The Meltdown attack specifically targets an Intel bug that fails to check uncommitted instructions for access violations before they can cause side effects (like cache loads). This is not the only way that the technique can be applied: a related attack was released by the same team, called Spectre, that causes side effects (like cache loads) using speculative execution and branch prediction. Most processors today use these techniques, so they are also vulnerable to the same family of attacks.

The Meltdown attack is more serious because it allows kernel and physical memory to be read by the attacker's code. Spectre merely allows access to memory within the same process as the attacker, which is primarily a problem for processes that run untrusted JIT code like web browsers.

For Intel to simply replace 10 years of its installed product is not enough. Switching to AMD is not enough. I think this may finally be the goad that makes people take new computer architectures like The Mill seriously.

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Thu Jan 04, 2018 1:25 am
by Trippynet
Raion-Fox wrote:I do not buy AMD because I cannot take the performance hit (Benchmarks of Ryzens show abysmal single thread performance compared to Haswell Xeons) and because I do not wish to support their poor quality chips.


I have to confess to being a bit perplexed by this. First of all, it's rather unfair to compare a consumer CPU with a professional/server grade CPU that costs twice as much. Comparing instead to Kaby Lake Core i7s, a Ryzen 7 1800X is only marginally behind a Core i7 7700K in single threaded performance with CPU-Z, and is faster than an i7 6600K at single-thread performance (and this is Skylake, not Haswell). Whilst I'd agree that Intel's latest chips are therefore faster at single threaded performance overall, I'd hardly call Ryzen's single core performance "abysmal". Multi-threading is increasingly important these days, and Ryzen does score very well here (my current system absolutely motors through video encoding with Handbrake for example).

Secondly, your claim of "poor quality chips" - how exactly? I presume you're aware that the Meltdown flaw affects Intel chips but not AMD ones? In this case, I'd put that down as a significant black mark against the quality of Intel's chips, rather than AMD.

Personally, I take a neutral approach - I've owned plenty of AMD and Intel systems over the years and I wouldn't call either of them "poor quality". Both companies have had their ups and downs over the years, but good competition is healthy for the industry.

Re: Intel backdoor: not a bug, a feature (buy a new processor!!!)

Posted: Thu Jan 04, 2018 3:20 am
by Irinikus
Trippynet wrote:Personally, I take a neutral approach - I've owned plenty of AMD and Intel systems over the years and I wouldn't call either of them "poor quality". Both companies have had their ups and downs over the years, but good competition is healthy for the industry.


Agreed, it doesn't help to be a fanboy of any particular brand, if they drop the ball, drop them.

Buy whichever chip best suits your needs.