dram is proven exploitable !!!!!!

http://arstechnica.com/security/2015/03/cutting-edge-hack-gives-super-user-status-by-exploiting-dram-weakness/

It's hardware all the way down.

You can't trust hardware anymore.
I haven't used the stuff in years.

The NSA hacked my nervous system so I became a transcendant being.

>>3
Yeah. Real programmers abstract the hardware in their minds as perfect computers that never have bugs and masturbate to mental constructs of their own creation.
>>4
Is your name ANDRU by chance?

>>1
All abstractions eventually have to deal with physical reality.
Our minds are biological brain interfaces(even if we have souls, the souls have to interface into physical downstream output).
CPUs are built on analog microelectronics(with a digital representation defining 0's and 1's).
The Internet is just cables and sattelites, it can be cut,interfered with or shut off. Dealing with reality is hard, but its the truth behind all abstractions and autistic escapism cannot nullify it.
Living in abstraction land where perfect RAM doesn't get corrupted and CPUs never have bugs is a recipe for disaster.
I hope this exploit is not applicable to JavaScript/Flash/Silverlight as this would create some serious botnet potential.

Kanter, of the Microprocessor Report, said. "If you want to actually fix this problem, we need to go out and replace, on a DIMM by DIMM basis, billions of dollars' worth of DRAM. From a practical standpoint that's not ever going to happen."

This is what happens when idiots start making decisions.

"Higher density! Lower power consumption! Lower cost!"

They get so caught up chasing after those brillant[sic] marketing features that they forget about the entire purpose of memory in the first place. If you read the original paper, you'll notice that this doesn't happen with memory older than ~2010.

Time to stock up on NOS DDR3 and older, I guess...

>>6
If Intel could offer free replacements for all the Pentiums affected by FDIV bug, there shouldn't be any reason why DRAM manufacturers today couldn't.

>>7
Nope, this would create a PR nightmare and huge costs(dram itself, shipping, legal work,etc). I bet they'll assume it'll just go away since the attack vector is so exotic and specific.

>>8
Clflush isn't exotic, actually you can simulate a flush by making cpu discard the cache, just allocate a huge array and write shit in every N bytes which invalidates the entire array, forcing the CPU to load it in the cache again and again and again.

>>9
Something like this?
//cachetest.c
#include "void.h" //gist.github.com/FrozenVoid/87e6ad6212ac9ce496e0#file-void-h"


#define cachelinesize 64
#define invalidate(x) sm for(u8 i=0,imax=elems(x);i<imax;i+=cachelinesize/sizeof(x[0])){x[i]++;x[i]+=x[i];}; em
//link -lpthread
vp cacheflush(vp c){u8p arr=malloc(8000000);forever{invalidate(arr);};re 0;}
vp somekindofcode(vp v){;re 0;}
STDSTART
pthread_t threads[17];;
//16 threads invalidate cache at same time
for(u4 c=0;c<16;c++)
pthread_create(&threads[c], NULL, cacheflush,NULL);
//extra code to run
pthread_create(&threads[16], NULL, somekindofcode,NULL);
for(u4 c=0;c<17;c++) pthread_join( threads[c], NULL);

STDEND


>>10

cache gang rape

>>7
When you're talking about hardware, you're actually making sense. Just don't wander into that "software" corner where you don't know shit.

>>11
Whom are you quoting?

Cudder is all talk and no action.

>>6
bullshiting so hard no i'm not gonna argue with baseless assertions by a confused ignorant nigger who cant even get his terminology right

>>7
No one is going to fix this on the DRAM side. The only long term fix is to add still more complex refresh logic to the controllers. Yes, continuing to scale past the point that the silicon will work reliably is insane, but everyone in JEDEC must have known the DRAM makers would do it anyway. The market pressure to do it is just too great.

If controller designers had begun work to incorporate features like PARA into their products in 2010, they would have it in chips that are shipping now. Nobody is saying anything to that effect publically so I'm going to take a wild guess that everybody missed their intercept.

Error: This thread is too old. You can't reply anymore.

Error: This thread has over 18 replies. You can't reply anymore.

>>18
Wrong. You're post is the 17^th reply. >>1 is not a reply. Something that always bothered me about Shii_tchan was how the bump limiter would make the Over 1000 Replies post right after the 998^th reply.

Error: This girl is too old. This girl is over 12 years old. You can't like her anymore.

>>10,11
Uhg, I can't stop laughing. What is wrong with me.

This thread has been closed and replaced with the following thread:

Subject: Compiling Photoshop code with the GIMP
Name:
Email:

It doesn't work.

>>22

It doesn't work.

Your dubs do, though.

>>14
Cudder takes no action because Cudder is subject to non-compete.