Open rgacogne opened 7 months ago
This sounds like a lot of work for little to no gain; is the concern here that libsystemd
links to many other libraries and one of those libraries could be used to compromise auth/rec/dnsdist?
Yes, and I agree it might not be worth it, I just want to make sure we know why we do it. I know systemd
is working on not loading the compression libraries unless they are actually used, which will make the situation better, but the libraries would still show up in the transitive dependencies of our products.
Systemd apparently did some efforts on cleaning things up in that regard: https://chaos.social/@bluca@fosstodon.org/112208167240646037
Also https://chaos.social/@pid_eins@mastodon.social/112202687973228580 might be of interest 😊
Short description
On Linux systems where
libsystemd
is available all three products automatically link against it. Given the recentxz
drama, I'm wondering if it would make sense to re-implement the communication primitives we need and drop this dependency, especially since the interface has been stable for a long time. For auth and dnsdist this means re-implementingsd_notify
which is easy. For the recursor (and possibly auth and dnsdist in the near future) it meanssd_notify
,sd_journal_sendv
andsd_journal_stream_fd
, which seems significantly harder. It might not make sense in the end, but I feel it's worth looking into to understand the trade-off.