[ruby-core:101182] Re: Spectre Mitigations

If you really do want to add barriers to your code, a one-line C extension with a full barrier instruction in it would be easy to write and would be the only way at the moment.

Chris

> On 1 Dec 2020, at 16:05, Amel <[e-mail]> wrote:
> 
> Trusted code can potentially be vulnerable to Spectre attacks, as branch prediction on modern processors occurs in all code where there are branches. The question is how can I stop security critical branches from being executed out-of-order and thus protect my trusted code?
> 
> Amel
> 
> On 01.12.20 16:44, Chris Seaton wrote:
>> I wouldn’t recommend using Ruby to run in-process untrusted code in the first place. Are people doing that?
>> 
>> Chris
>> 
>>> On 1 Dec 2020, at 15:32, Amel <[e-mail]> wrote:
>>> 
>>> Hi there!
>>> 
>>> I've already asked this question in the ruby-talk mailing list and it was hinted at that ruby-core could be the better place to ask, so here's my question.
>>> 
>>> Are there any mitigations in Ruby for the Spectre security vulnerability? Maybe in the interpreter itself or a function which is similar to the lfence-instruction.
>>> If there isn't, are there any mitigations planned?
>>> 
>>> Cheers,
>>> 
>>> Amel
>>> 
>>> 
>>> Unsubscribe: <mailto:[e-mail]?subject=unsubscribe>
>>> <http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>
>> 
>> Unsubscribe: <mailto:[e-mail]?subject=unsubscribe>
>> <http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>
> 
> Unsubscribe: <mailto:[e-mail]?subject=unsubscribe>
> <http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>


Unsubscribe: <mailto:[e-mail]?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>