Re: Are memory safety and maximized data volatility mutually exclusive?

Not sure if the original got through. Sent from wrong address. :-)

On Thursday, October 8, 2015, Jonathan S. Shapiro <
[email protected]> wrote:

> On Thursday, October 8, 2015, Scott Moore <[email protected]
> <javascript:_e(%7B%7D,'cvml','[email protected]');>> wrote:
>>
>> The language standards are quite nice about letting compilers elide code
>> that looks like make-work (why write a zero to this memory if no one will
>> read it in the future?).
>>
>
> It's true that compilers can optimize this way, but only when the compiler
> can *witness* that the write-zero is never consumed. Performing the zero
> operation in a separately compiled procedure that is **NOT** specified by
> the standard library (and therefore not elidable by the compiler based on
> knowledge of the library specification) is generally sufficient. The
> zeroing procedure will be conservative about use-after-store, and the call
> to the zeroing procedure in a separately compiled object file cannot be
> eluded because the compiler does not know in general what important side
> effects may later occur.
>
> Though strange things can happen if the call to a side effecting procedure
> is immediately followed by exit()...
>
> Jonathan
>

_______________________________________________
cap-talk mailing list
[email protected]
http://www.eros-os.org/mailman/listinfo/cap-talk