Bug#776799: unblock: systemd/215-11
Control: tags -1 d-i
On 2015-02-01 21:16, Martin Pitt wrote:
> Package: release.debian.org
> Severity: normal
> User: release.debian.org@packages.debian.org
> Usertags: unblock
>
> systemd 215-11 hit unstable three days ago with some important/safe
> fixes. So far there have been no regression reports.
>
> I attach the full debdiff between 215-10 and -11, but as usual I also
> link to the individual commits on anonscm. Note that there are zero
> changes for udev and hence the udebs (for d-i).
>
Hi,
Thanks, looks good to me.
As a formality, it still needs a d-i ack for this. :)
The rest of the message is quoted in full for KiBi's convenience.
Thanks,
~Niels
> Annotated changelog:
>
> | [ Martin Pitt ]
> | * escape-beef-up-new-systemd-escape-tool.patch: Avoid creating a danling
> | symlink, to work around regression in recent patch (see #776257).
>
> That's an RC bug in patch, but easily worked around in systemd, so
> that we can actually upload new versions until it gets fixed. patch
> still needs to be fixed properly to unbreak glibc and a bunch of other
> affected packages, of course.
>
> http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=492416
>
> Sorry for the typo in the changelog! (Fixed in git).
>
> | * Order ifup@.service and networking.service after network-pre.target.
> | (Closes: #766938)
>
> network-pre.target in Debian's shipped units only occurs in one
> package [1]; so this will make no practical difference on most systems,
> but will make ifup@.service actually behave according to documentation
> and fix pyroman's startup order to really start the firewall before
> bringing up network interfaces.
>
> http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=b29acf
>
> [1] http://codesearch.debian.net/results/network-pre.target
>
> | * Tone down "Network interface NamePolicy= disabled on kernel commandline,
> | ignoring" info message to debug, as we expect this while we disable
> | net.ifnames by default. (Closes: #762101, LP: #1411992)
>
> Confusing, and trivial.
>
> http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=9cc4d0745
>
> | * logind: handle closing sessions over daemon restarts. (Closes: #759515,
> | LP: #1415104)
>
> This is admittedly a bit intrusive, but still relatively
> straightforward; I'm happy to answer further questions about it. I've
> heard this error (also on Ubuntu) from a lot of users by now, and it
> essentially breaks your running desktop sessions on installing package
> upgrades so I'd really like to fix this for Jessie. I have tested it
> fairly thoroughly.
>
> If you object or have doubts, I'm ok with backing this out again in
> -12.
>
> http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=37ce19d
>
> | * logind: Fix sd_eviocrevoke ioctl call, to make forced input device release
> | after log out actually work.
>
> This is quite an obvious fix of the ioctl; calling this "security" is
> a stretch as we don't really use this feature in Debian yet. But it's
> an interesting thing if you have a multi-seat system and use Wayland
> instead of X.
>
> Again, if you object I'd be okay with backing this out if you object.
>
> http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=2d0f077
>
> | * debian/patches/series: Move upstreamed patches into the appropriate
> | section.
>
> No actual run-time effect, this just clarifies the status of our
> patches.
>
> http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=0d5964
>
> | [ Michael Biebl ]
> | * Make sure we run debian-fixup.service after /var has been mounted if /var
> | is on a separate partition. Otherwise we might end up creating the
> | /var/lock and /var/run symlink in the underlying root filesystem.
> | (Closes: #768644)
>
> http://anonscm.debian.org/cgit/pkg-systemd/systemd.git/commit/?id=a09ebde9
>
> Trivial and obvious fix; that bug can break your system pretty badly
> if you hit the race.
>
> Thanks for considering!
>
> Martin
>
> unblock systemd/215-11
>
Reply to: