The new interruption, however, is separated for now CVE-2020-0551 e Intel-SA-00334, discovered by the same team of researchers that found the popular Meltdown and Specter, and actually used the same process to extract sensitive data through a processor: in a type of attack called Weight Number Needle (LVI).
Adjusting the risk will have a significant impact on performance
A malicious attacker can take advantage of microarchitecture data leaks from Intel processors to install and encrypt malicious code in a way that violates privacy storage systems, thus being able to extract, in fact, as much data as it wants from a computer.
According to investigators, Intel processors are already full of risk-averse solutions found two years ago, so resolving these vulnerabilities would mean changing all processors' computations, and during the initial testing they performed a full reduction determined that operating losses range from 2X to 19X, which pays off in a way that brings back nearly what Intel processors have become – in terms of performance – almost 10 years ago.
This is because LVI protection includes compiler and compiler integration that includes additional x86 commands (lfence), replacing vulnerable (ret ret) commands with sequential but detailed data. In other words, because problem solving you will need to change the instruction sets of the processor, making it slow because the rest of the command will be more difficult to execute.
Another danger of being Intel
How many have already gone? The situation is worse. If the company's reputation has already been in question for two years because new threats affecting its competitors continue to emerge, the one we are talking about today could be the last resort because a reduction in performance of such magnitude is unacceptable, and it is clear that Intel will not fix it this way.
So the options are left by the two biggest silicon giants: leave the risk open by accident involving, or they need another way to fix it because, as we said, the loss of functionality that separates between 2X and 19X is unacceptable and they cannot afford it. In his defense it must be stated that this reduction in the most effective is the recommendation of the investigating team and should not be the only solution. Of course Intel will work on it.
