Potential issue reported from rs-matter

[skoky]

Hi

Please have a look at discussion in project-chip/rs-matter#460 and problem report from ivmarkov. He exaplain potential issue in rust-matc. I am not sure I understand it enough to validate.

[andresv]

I think these:

https://github.com/tom-code/rust-matc/blob/main/src/session.rs#L26
https://github.com/tom-code/rust-matc/blob/main/src/device/mod.rs#L94

Should be masked like this:

AtomicU32::new(rand::random::<u32>() & 0x0fff_ffff)

Allegedly spec (§4.6.1.1) says (Haven't checked by myself)

All message counters SHALL be initialized with a random value using the Crypto_DRBG(len = 28) + 1 primitive. The random initializer ranges from 1 to 2²⁸.

[tom-code]

yes, specification says that. thanks.