Re: [PATCH] Log details for client certificate failures

Hi,

On 2022-07-15 13:20:59 -0700, Jacob Champion wrote:
> On 7/15/22 12:11, Andres Freund wrote:
> > This might have been discussed somewhere, but I'm worried about emitting
> > unescaped data from pre-auth clients. What guarantees that subject / issuer
> > name only contain printable ascii-chars? Printing terminal control chars or
> > such would not be great, nor would splitting a string at a multi-boundary.
> 
> Hm. The last time I asked about that, Magnus pointed out that we reflect
> port->user_name as-is [1], so I kind of stopped worrying about it.

I think we need to fix a number of these. But no, I don't think we should just
add more because we've not been careful in a bunch of other places.


> Is this more dangerous than that?

Hard to say.


> (And do we want to fix it now, regardless?)

Yes.


> What guarantees are we supposed to be making for log encoding?

I don't know, but I don't think not caring at all is a good
option. Particularly for unauthenticated data I'd say that escaping everything
but printable ascii chars is a sensible approach.

Greetings,

Andres Freund