Detecting a misremembered passphrase in gpg-agent

[Jack]

On 2024.06.13 06:57, ael via Gnupg-users wrote:
> Further thoughts on detecting a mistaken passphrase entry when
> encrypting. I have looked at both
>   man gpg-agent  and info
> and I could not immediately see anything to help, but I quickly became
> lost in the overwhelming volume of the entries :-)
> So perhaps there is something there that I have missed.
> 
> The user case is not the "usual" use of gpg for communicating with 2nd
> parties. Rather I am using symmetric encryption on local files and
> usually using a common (long) passphrase on a common set of those
> files. The plain text file is deleted after encryption for
> security. So if I make a mistake in entering the passphrase I have  
> lost
> access.
> 
> pinentry asks me to repeat the passphrase and that is obviously the
> main defence against getting things wrong. However, I am quite capable
> of misremembering a component of a passphrase that I have not used  
> for a
> long time, or even using the wrong passphrase in an absent-minded
> moment.
> 
> Having to repeat a long passphrase is quite laborious, and the
> suggestion below would solve that.
> 
> My simple suggestion is that there be an option, perhaps even a  
> tick-box
> on the entry window, that displays a checksum/fingerprint/hash of the
> entered passphrase. That hash can then be checked perhaps manually,
> perhaps directly against the known hash of the passphrase. If it is
> checked manually, it needs to be quite short. If the hash matches,  
> there
> is no need to re-enter the passphrase. It also guards against
> re-entering a misremembered phrase.
> 
> Something like this would be a huge improvement for my use case.
> Probably useful more generally. Of course, you would still need double
> entry when initially setting up a passphrase which does not yet have a
> hash.
> 
> ael
I'm no expert in this area, but something struck me - is the passprase  
you are entering protecting the key you are using for encryption, or is  
the passphrase itself being used for encryption?  From your  
description, it seems like the latter.  If you had created a key for  
the encryption, and then protected that key with the passphrase, then  
mistyping the passphrase would just get you a failure and not a  
successful encryption with the wrong key.

Does this help at all, or have I missed something?

Jack