Christian Noé Ramos López reported two problems that, taken together, mean an SSH signing key cannot be reliably stopped from being trusted. He tested git 2.47.3 with OpenSSH 10.0p2 on Debian 13, and read the source at v2.47.3 and at master.
The first problem concerns valid-before. SSH signatures carry no time of their own, so Git passes -Overify-time from the committer or tagger line. That date is written by the signer. In his test, alice's key was in the allowed signers file with valid-before="20260101". A commit with a backdated committer line (June 2025) verified as good (%G?=G), while an honestly dated commit from October 2026 showed U and was rejected by verify-commit and merge checks.
The second problem is about gpg.ssh.revocationFile. When it names a file that does not exist, Git warns and runs ssh-keygen without it, so a signature made with a key listed in the intended revocation file verifies as good. A revocation file that exists but cannot be read already makes ssh-keygen refuse. Christian also notes that sshd refuses all public-key authentication when its RevokedKeys file is unreadable.
His patch fails verification when the revocation file is missing and turns the warning into an error so the reason is shown. It does this in the block that already fails keys not in the allowed signers file, so a key that is in that file still gets its signature info displayed and %G? gives U in both cases.
In the latest message, Christian says he followed Junio C Hamano's review comments. v3 replaces an earlier comment with one at the top of the block saying check-novalidate runs there only to display the signature info. He also noted that v2 was mistakenly sent as a reply to the wrong message and that v3 will be a reply to v1. The excerpts do not show any proposed change for the backdating problem.