One of OCSP's problems is the dominance of Chrome

May 9, 2024

To simplify greatly, OCSP is a set of ways to check whether or not a (public) TLS certificate has been revoked. It's most commonly considered in the context of web sites and things that talk to them. Today I had yet another problem because something was trying to check the OCSP status of a website and it didn't work. I'm sure there's a variety of contributing factors to this, but it struck me that one of them is that Chrome, the dominant browser, doesn't do OCSP checks.

If you break the dominant browser, people notice and fix it; indeed, people prioritize testing against the dominant browser and making sure that things are going to work before you put them in production. But if something is not supported in the dominant browser, it's much less noticeable if it breaks. And if something breaks in a way that doesn't affect even less well used browsers (like Firefox), the odds of it being noticed are even lower. Something in the broad network environment broke OCSP for wget, but perhaps not for browsers? Good luck having that noticed, much less fixed.

Of course this leads to a spiral. When people run into OCSP problems on less common platforms, they can either try to diagnose and fix the problem (if fixing it is even within their power), or they can bypass or disable OCSP. Often they'll chose the latter (as I did), at which point they increase the number of non-OCSP people in the world and so further reduce the chances of OCSP problems being noticed and fixed. For instance, I couldn't cross-check the OCSP situation with Firefox, because I'd long ago disabled OCSP in Firefox after it caused me problems there.

I don't have any particular solutions, and since I consider OCSP to basically be a failure in practice I'm not too troubled by the problem, at least for OCSP.

PS: In this specific situation, OCSP was vanishingly unlikely to actually be telling me that there was a real security problem. If Github had to revoke any of its web certificates due to them being compromised, I'm sure I would have heard about it because it would be very big news.

Written on 09 May 2024.
« All configuration files should support some form of file inclusion
It's very difficult to tell if a Linux kernel bug is a security issue »

Page tools: View Source, Add Comment.
Login: Password:
Atom Syndication: Recent Comments.

Last modified: Thu May 9 23:23:35 2024
This dinky wiki is brought to you by the Insane Hackers Guild, Python sub-branch.