In two years from now, look at which certificate authorities and how that affects the pins we have in Mozilla Firefox and Google Chrome, see #41154 (closed) for background and the previous instance of this.
Designs
Child items 0
Show closed items
No child items are currently assigned. Use child items to break down this issue into smaller parts.
The reason we're considering removing them is we've not been great at keeping track of the pins, and what's happening today is not unlikely to happen in the future.
I'm also wondering now that we have proper CAA records and have made progress in monitoring CT logs, whether these pins really add much to the security of users who navigate to these handful of websites covered.
Another major reason is that it seems nowadays most vendors are recommending against certificate pinning because of the high risk compared to the low benefit it brings:
This file in chromium has a base64 encoded set of certificates, there are very few in here, so it is remarkable to find:
Tor1-----BEGIN PUBLIC KEY-----MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAlOpp8zfF+jEbI6R7nxnmphFPqBCCHMpahBMjQOjHD/860wrH1ZoIeevW9ZJ6MIuBV0jwwtAQokNsx8FVAmMHTjg6b4blbum1meahODgafcGjHrRGzEvuN6r577SgHcOQXkAIVsbZ6Bjk+8kh1t/vcEaLa9DmzBQpMTYWCDY4kN0tFMSNS8CJekgA6vDVFHFRp4NiJ1eFmd38furKsZUZrqvY4jzyaIls/UQB8q/6bHbEeF5lTUaWPZob1XBv0Ib7nVX2gGzan+qYQim8WdjcLrGhnUBwil5lh9qQ/8VIMVbjtcAjeZpVWuTZA3stbOiaDCqJBGTe9glFOH1xz9hS7QIDAQAB-----END PUBLIC KEY-----Tor2-----BEGIN PUBLIC KEY-----MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEA1b2qGT7y4sjNG18lZKuCtLwDfE57KseaNY6VM61DySaobdDt9VOiFS+3QaPcrT60qOH9aQ1DGhNv3jcAJvtTi5nKSgVDu7Emb70j2xrRaJ+lnITYs2Je9JYUQmb6IQAplYTz60s0Ng1MrROqNswlUhhXTCZLB1Bo+uN8OM7LoZ+Sr+sP0YhMsNAyxf3ZJVvzhWHQYGHvNBqIrXpq6YLczHSdd5QirZ3lSdhv2uGYxDBLslfk47xObxSvqblLD2ISlMRPL85FnetHL/dRMUEo2c9MJ49nxItUbeizxxIeCbfWdbucd4il5bBlN9t3duFqwCu80ZOuHMvxavpP05D5RwIDAQAB-----END PUBLIC KEY-----Tor3-----BEGIN PUBLIC KEY-----MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAtNfM24/03NeOb3R1sRN57zwYThdBJPSOXehVN3JvShTZ86kg+f0HYwbZZ0Xx0jujghabKgHkhiLMChl+Fg97Pi0t4mQm/3sdltfUypprdcE6CbCzKkWNTaD9+iMA4m8peSq0Pd5hCKvRidb1+OCLrJKOqx24r9VwsIa69zdiE5DpFBqvi1YczoaknF09jNR/hjqMyEwENR7t/AKZ1VpK6L6loJt2mCl3FWjM+Wt50ktrSyAI4aPfSAILAmSRhb5TY1xWRBuHDOzdf+DvegXu/aHn0j5UscN2XqQiNJcUrdJYsekfpHDtYKP/pesHCp2lCSbYdAaS8mnqDJwV8xC41wIDAQAB-----END PUBLIC KEY-----
I don't understand how these are used, or how they are linked to tor itself, or how we go about getting them removed.
Yep, I just found that as well. I think it may be the place.
This suggests that we are going to have this same problem with Chromium with the newer LE cert intermediaries, and will need to get a cert from one of the EV providers in that list until an updated version of this list has trickled down to users.
Google responding and will be removing the pins. They said that the source of truth for these is an internal file and the source files that we have been linking to here are updated on a daily basis based on the contents of that internal file. That means we will see an update tomorrow or the day after depending on when the internal change lands. Additionally, most Chrome users will also get a component update with the change, so it will take effect even before it's visible on the source files (and will reach users with old versions of Chrome who don't update).
We decided to remove the pinnings and have confirmed this with Mozilla and Google which have both acted on our request.
In #41672 (closed) we'll be following up about ensuring the ~3 month period required for pinning expiration is handled gracefully for users of older browsers.