From 4f8a648cc44d87b8a8e4c94e16e87aff577fe4f9 Mon Sep 17 00:00:00 2001 From: Daniel Krol Date: Fri, 22 Jul 2022 16:37:27 -0400 Subject: [PATCH] Notes for auditor. --- server/client.go | 6 ++++++ test_client/test_client.py | 5 +++++ 2 files changed, 11 insertions(+) diff --git a/server/client.go b/server/client.go index 2424176..22e0a2b 100644 --- a/server/client.go +++ b/server/client.go @@ -12,6 +12,12 @@ import ( // Thanks to Standard Notes. See: // https://docs.standardnotes.com/specification/encryption/ +// +// Auditor: I don't really understand how this system exactly works, and if +// I'm doing it right here, given that I don't understand it. In particular: +// Email address isn't sufficient for a secure salt, but it *is* somehow +// sufficient to keep the server from lying to us about passing the seed +// between clients? Is that the idea? type ClientSaltSeedResponse struct { ClientSaltSeed auth.ClientSaltSeed `json:"clientSaltSeed"` diff --git a/test_client/test_client.py b/test_client/test_client.py index 0c140cc..a12e122 100755 --- a/test_client/test_client.py +++ b/test_client/test_client.py @@ -318,6 +318,10 @@ class Client(): def register(self): # Note that for each registration, i.e. for each domain, we generate a # different salt seed. + # + # Auditor - Does changing salt seed here cover the threat of sync servers + # guessing the password of the same user on another sync server? It should + # be a new seed if it's a new server. self.salt_seed = generate_salt_seed() self.lbry_id_password, self.sync_password, self.hmac_key = derive_secrets( @@ -514,6 +518,7 @@ class Client(): # update that as well so that the sync password and hmac key are derived # from the same root password as the lbry id password. + # Auditor - Should we be generating a *new* seed for every password change? self.salt_seed = generate_salt_seed() new_lbry_id_password, new_sync_password, new_hmac_key = derive_secrets( new_root_password, self.email, self.salt_seed)