Bug#1105036: openssh: please produce OpenSSH 10.x backports for Bookworm and Bullseye
Colin Watson <[email protected]>
| Newsgroups | gmane.linux.debian.devel.ssh |
|---|---|
| Message-ID | <aCs6SAITcPOvzmSP__46011.7491705873$1747663587$gmane$org@riva.ucam.org> |
On Mon, May 19, 2025 at 11:45:40AM +0300, Martin-Éric Racine wrote: >su 11.5.2025 klo 18.33 Martin-Éric Racine ([email protected]) kirjoitti: >> su 11.5.2025 klo 17.17 Colin Watson ([email protected]) kirjoitti: >> > On Sat, May 10, 2025 at 12:34:34PM +0300, Martin-Éric Racine wrote: >> > >Given how Trixie has raised the minimum CPU level for some >> > >architectures (e.g. on i386, minimum Pentium 4 due to Rust now >> > >requiring SSE2), it would be desirable for OpenSSH to be backported to >> > >at least Bookworm, preferably also to Bullseye, so that hosts running >> > >on deprecated CPUs can remain reachable using post-quantum crypto for >> > >the remainder of their LTS support timespan. >> > >> > While I'm reasonably OK with doing this, backports rules mean that I >> > need to wait until there's a suitable version in testing, and I'd much >> > rather wait until at least version 1:10.0p1-3 is in testing since the >> > current version in testing has an RC bug. >> >> That seems reasonable enough. Looking forward to it. > >Can we return to this now that we have 1:10.0p1-5 in Testing? Thanks for the reminder. I've uploaded 1:10.0p1-5~bpo12+1 to bookworm-backports. I did try bullseye, but the regression test suite hung at "conch ciphers: cipher aes256-ctr". That probably just needs reintroducing an old patch to deal with older versions of Twisted, but this sort of thing can get quite time-consuming and I'm reluctant to spend a lot of time on it with weak justification. I propose to just do bookworm and call that good enough; users of affected CPUs can at least upgrade to bookworm.