Hi,
I've received a bug report saying that Nettle functions leave secret
values lying around on the stack or in deallocated heap memory when they
return.
Nettle generally does not take any steps to wipe or cleanup memory that
might hold secrets when no longer used. Which I think differs from many
other crypto libraries.
In the past, I have argued that an application can't really know which
parts of its user data is secret and which isn't. Not only cryptographic
keys are sensitive, the plaintext protected by those keys might be as
sensitive. And sensitive data is typically handled also by
non-cryptographic applications, say your text editor or email program or
ai assistant. For this reason, it ought to be the responsibility of the
operating system to protect each application's memory, with techniques
like MMU-based memory protection, and encryption of any memory pages
swapped out to disk.
So my question: Would it improve security if Nettle is changed to make
more of an effort of explicitly clear deallocated (or otherwise no
longer used) memory on the heap and the stack? Or maybe make an mainly
regarding the stack; an applications can configure custom heap
allocation functions, and that way ensure that memory is cleared on
free, but there's no corresponding application control over Nettle's
stack usage.
Regards,
/Niels
--
Niels Möller. PGP key CB4962D070D77D7FCB8BA36271D8F1FF368C6677.
Internet email is subject to wholesale government surveillance.