| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001496 | Main CAcert Website | GPG/PGP | public | 2020-10-31 13:48 | 2024-10-05 07:40 |
| Reporter | NoSubstitute | Assigned To | Ted | ||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | needs review & testing | Resolution | open | ||
| Platform | Main CAcert Website | OS | N/A | OS Version | stable |
| Summary | 0001496: CAcert signed GPG key reports Invalid Digest Algorithm | ||||
| Description | I'm guessing this could be why. gpg: Note: third-party key signatures using the SHA1 algorithm are rejected sig% P X 0xD2BB0D0165D0FD58 2019-09-22 [Invalid digest algorithm] sig% P 0xD2BB0D0165D0FD58 2020-10-31 [Invalid digest algorithm] >gpg --version gpg (GnuPG) 2.2.23 libgcrypt 1.8.6 Copyright (C) 2020 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <https://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Home: C:/Users/Kim/AppData/Roaming/gnupg Supported algorithms: Pubkey: RSA, ELG, DSA, ECDH, ECDSA, EDDSA Cipher: IDEA, 3DES, CAST5, BLOWFISH, AES, AES192, AES256, TWOFISH, CAMELLIA128, CAMELLIA192, CAMELLIA256 Hash: SHA1, RIPEMD160, SHA256, SHA384, SHA512, SHA224 Compression: Uncompressed, ZIP, ZLIB, BZIP2 | ||||
| Steps To Reproduce | Sign a gpg key with the gpg signing feature of cacert. Import key to gpg. Check the signatures of the key. | ||||
| Additional Information | Patch created in https://bugs.cacert.org/view.php?id=1473 | ||||
| Tags | No tags attached. | ||||
| Reviewed by | |||||
| Test Instructions | |||||
|
|
This is because SHA-1 signatures are unsafe and can be abused in practice: https://sha-mbles.github.io/ GnuPG has been rejecting them for more than one year. I opened a bug report about one year ago (0001473) to warn CA Cert before making the attack public, but nothing has happened since and CA Cert still uses SHA-1 to certify user's PGP keys. This is really a shame. |
|
|
Yes, this should have been solved by now. How about you two, gleurent and NoSubstitute, take care of this case? For our volunteer software team, there would still be 1000 bugs left... |
|
|
I wish I had any coding skills to offer. I can help out testing if someone else does the coding part. :-) |
|
|
@NoSubstitute coding is already done. It just needs to be tested. The patch has been applied on test.cacert.org since May 17th 2020. |
|
|
@jandd alright. I just registered on the test system, as I didn't have an account there. I'm not getting the email verification. |
|
|
@NoSubstitute thanks for volunteering for tests. The test system sends its mails to a special catchall mailbox and does not send it to the Internet. You can go to the test manager at https://test.cacert.org:14843/mail and log in with your test system credentials. This gives you access to your mails from the test system. There is a function to give you assurance points at https://test.cacert.org:14843/manage-account/admin-increase. You need 100 points to be able to use the GPG function. |
|
|
Thank you. That part worked great. Managed to add points, and add secondary email, and request GPG signature. Grabbed signature, and imported to Kleopatra, and it doesn't complain about invalid algorithm for that signature. However, it says "no public key" (see attached image), as the test GPG key doesn't seem to be the same as the one for the PROD system, and I find no reference to where I can grab the TEST GPG key, so I can verify that the signature is fine. Also, the email sent contains references to the PROD GPG key, as well as a download URL on the PROD system, but with one unprintable character, as seen below. *** Hi Kim, Your CAcert signed key for kim.nilsson@no-substitute.com is available online at: https://www.cacert.org/gpg.php?id=3&cert054 To help improve the trust of CAcert in general, it's appreciated if you could also sign our key and upload it to a key server. Below is a copy of our primary key details: pub 1024D/65D0FD58 2003-07-11 CA Cert Signing Authority (Root CA) <gpg@cacert.org> Key fingerprint = A31D 4F81 EF4E BD07 B456 FA04 D2BB 0D01 65D0 FD58 Best regards CAcert.org Support! *** 2020-12-26 165239-CAcert-GPG_test-signature.png (5,896 bytes) |
|
|
@NoSubstitute thanks for testing. The test system uses a different gpg key: -----BEGIN PGP PUBLIC KEY BLOCK----- Version: GnuPG v1 mQGiBE3TAtQRBACuFhUNT3BOIXflZx8ENFN2hf0KHDgReWMW5+hcx8M2C9mueAST 89x0QxtdWUbsrMFSr3UvDIKjoPFVzV+WT/dez5aADSrnR4OxYPfzcNIWPjm8su/9 HrH6+ivaH+iOoBOuhYrHthun+1fsqLedhTMNqb+tt6OfzqIuyFkmpm+FqwCg5ECF s53E7j6BJuKYLeWqFDqFOH0D/1pnXyBF78YurXcwwA7eSQWoc3f1g5lwyqOWwcLv ynI/dcPV5BosEUcqEjaNEyBR0LFRZC7tnw2VLPzr65aLA6SrFqEJ2bEdWkO9vM5g 8InhmcqTz7Yg/QGnyGT0iNBMG8zg4Ajz20Ty6GE887HMao6qpYKXSbujMRWVjVLn VjpgA/9i1gsBjpGwr5rJ67dyDLVkJ5XMEP2ivgAhzKQ/inyLk2N3Nbm0YKWgrZRE +pEcZVsLn2jHqd5yu1zVr3Cm44HHaB/A4ew+hRio9vbhJy7cfanfZpAL8Rg8wZbQ t4x6zvrs0fXXNuD1BiSucOK4qGZ2vWukggmhAlHu5x5hws5qtbRLQ0FjZXJ0IFRl c3QgU2VydmVyIFNpZ25pbmcgQXV0aG9yaXR5IChmb3IgdGVzdGluZyBvbmx5ISEh KSA8Z3BnQGNhY2VydC5vcmc+iGYEExECACYFAk3TAtQCGwMFCSWYBgAGCwkIBwMC BBUCCAMEFgIDAQIeAQIXgAAKCRBL5zSBd/dRrLxrAJ9kaSLRAlcy9e57yuAAxoZy weX20ACg44B1AA26G0OePf9o3YBdUTjPsES5Ag0ETdMC1BAIAI9UEZ2NKbJfVWc+ CBb5kKhO6qazLDnRsRkMY5WE1PTBP09ieAaQ7cv424e6fKdZxSO5vUQ27tLmcQXo ICoREsDuSyESvEPbyBOro5LIY+z52TihqEHLPUN7OxlbPLX81AIKOCgEIUtpcYoX MiwXPVO50Swo3PCCQiOQ7ewvkd32ZbSPl50B+/59JiDsQsnx1o+Ls7WkALzQ6w7T UxLIVYWuFd9Hg/VOXBH67e81p/O7SDy6pMkDQaA9F22rA5Vv5TBJ2BAKWYfYK47i HxtUhw2/WBUzhcw2AGqCnCy9nrYELLAqljnlCdWLyvhuWyUAG87D/1/5qD0+9idh jcFWH/MAAwUH/2R7AWSZzbomxTcO5Q7/WxzTDtgUEy7eRbjqFMAusiQlTYDdeHz8 PfWok6mGhUapJaGnA05hy6aYBIzl4idmQgz39xD47O0qeDWve6YwrDuLYju6JIwv YfNiNAOjrZcRW4PCYeJfIK0WHV3+3kkneOX4Mql9QNKWs5W2LReMnHyq6POWasuf EgDu0TfqaJTeK3IXKrEJI8G4R4xdHXmZ16Is5T49w09mvMan+TTGczzjWn0aWD3D DOlTKK7h+82MSNCTiCi7aNoEyMTfeadSsrbqiaqtF0P/fyBDpGWFj3/A4foAcgtx KVt0UHf1Va1g8M+czG3pCOne6dyafZGAe7qITwQYEQIADwUCTdMC1AIbDAUJJZgG AAAKCRBL5zSBd/dRrAC5AKDdxp26zp4Km3jItBdeMD/gv8qxMgCgmmH1kizLe5Rk yYchS2oleEWXIqM= =9tD2 -----END PGP PUBLIC KEY BLOCK----- The mail template is hard coded into the signer client script. The special character in the mail link may be introduced by some encoding issue. The code line in CommModule/client.pl looks sane: $body .= "https://www.cacert.org/gpg.php?id=3&cert=$row{id}\n\n"; but maybe the Content-Transfer-Encoding: quoted-printable in the sendmail function in the same script breaks the content. I will file a separate bug for the mail template issues. |
|
|
With the test key imported the signature looks great. 2020-12-26 175400-CAcert-GPG_test-signature-valid.png (9,339 bytes) |
|
|
@Ted seems like the patch has been tested successfully. What are the next steps to get it in the production system (I know a site visit because it is code on the signer). Someone needs to merge the code into the proper branch and make a formal delivery to the critical team? Do we have a documentation of how this should be done? Can we add such a documentation to codedocs? |
|
|
@jandd, those were exactly the questions which discouraged me to continue work on this bug... The signer machine is something like a big black box to me. :-\ First of all the changes have to be reviewed by a Software Assessor, probably me. Where are these changes archived? I did not find a branch-1496 in the cacert-devel repository on github or git.cacert.org (is the signer sofware part of this repository at all? Only CommModule/server.pl?), is it somewhere in the svn repository? Or where else can I have a look at what was changed? The next question is whether the installation on the testserver is sufficiently simmilar to the signer installation. I did not find the signer machine on infradocs, am I looking at the wrong places? Maybe @egal can give some info here? Or to we have to ask Wytze? Without knowing the file layout on the signer it's quite hard to create some auditable install package. |
|
|
Awww, what a shame no progress was made on this now that there was a forced trip to the server. Created a new signature for my GPG key today, and it's also invalid; just like it has been since 2019. |
|
|
IMHO this is a duplicate of 0001473, we should continue work there. |
|
|
Should we close this, as it is indeed a duplicate of 0001473? Or should they both be closed when they are actually fixed? |
