HS Descriptors should only contain UTF-8
In #18938 (moved), we discovered some relays publishing platform lines or extrainfo documents with non-ASCII characters.
This seems to occur due to memory corruption.
So let's make sure that a non-ASCII UTF-8 check is done on:
- hidden services: unencrypted or encrypted descriptor
- HSDirs: encrypted descriptor
- clients: unencrypted or encrypted descriptor
See https://gitweb.torproject.org/torspec.git/tree/proposals/285-utf-8.txt#n45
This might require a spec change - it's likely implicit in the spec already.
For completeness, the set of printable ASCII characters is defined in dir-spec.txt ... Where I assume "printing ASCII" means "space to tilde", but we should also clarify that in the torspec.
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information