diff options
author | Earl Warren <109468362+earl-warren@users.noreply.github.com> | 2023-09-01 15:45:22 +0200 |
---|---|---|
committer | GitHub <noreply@github.com> | 2023-09-01 13:45:22 +0000 |
commit | 4ab8e56c9197135a85300542d736a556203752a6 (patch) | |
tree | bc49c3f49f0ca48245c8ad1af1cac74309a84e83 /webpack.config.js | |
parent | 9eb4a9e601c16174306d81ec4d73ffa5a0adc788 (diff) | |
download | gitea-4ab8e56c9197135a85300542d736a556203752a6.tar.gz gitea-4ab8e56c9197135a85300542d736a556203752a6.zip |
restrict certificate type for builtin SSH server (#26789)
- While doing some sanity checks over OpenSSH's code for how they handle
certificates authentication. I stumbled on an condition that checks the
certificate type is really an user certificate on the server-side
authentication. This checks seems to be a formality and just for the
sake of good domain seperation, because an user and host certificate
don't differ in their generation, verification or flags that can be
included.
- Add this check to the builtin SSH server to stay close to the
unwritten SSH specification.
- This is an breaking change for setups where the builtin SSH server is
being used and for some reason host certificates were being used for
authentication.
-
(cherry picked from commit de35b141b79a3d6efe2127ed2c73fd481515e481)
Refs: https://codeberg.org/forgejo/forgejo/pulls/1172
## :warning: BREAKING :warning:
Like OpenSSH, the built-in SSH server will now only accept SSH user
certificates, not server certificates.
Co-authored-by: Gusted <postmaster@gusted.xyz>
Co-authored-by: Giteabot <teabot@gitea.io>
Diffstat (limited to 'webpack.config.js')
0 files changed, 0 insertions, 0 deletions