systemd.git
7 years agoDo not apply uaccess tag for /dev/kvm if mode is 0666
Michael Biebl [Fri, 13 Jul 2018 21:36:13 +0000 (23:36 +0200)]
Do not apply uaccess tag for /dev/kvm if mode is 0666

(cherry picked from commit ace5e3111c0b8d8bfd84b32f2c689b0a4d92c061)

Gbp-Pq: Name Do-not-apply-uaccess-tag-for-dev-kvm-if-mode-is-0666.patch

7 years agoRe-add uaccess tag for /dev/kvm
Michael Biebl [Tue, 10 Jul 2018 12:47:58 +0000 (14:47 +0200)]
Re-add uaccess tag for /dev/kvm

If --dev-kvm-mode is set to something different then 0666, which we
explicitly support, it makes sense to still apply the uaccess tag to
/dev/kvm. For distros which opt to use the default 0666, this change is
a nop.

This partially reverts commit b8fd3d82205f632ce001fade74fed287e1564a1a.

(cherry picked from commit fa53e24130af3a389573acb9585eadbf7192955f)

Gbp-Pq: Name Re-add-uaccess-tag-for-dev-kvm.patch

7 years agonetwork/link: Fix logic error in matching devices by MAC
Dave Reisner [Sat, 7 Jul 2018 09:39:01 +0000 (05:39 -0400)]
network/link: Fix logic error in matching devices by MAC

Prior to this commit, a .link file with a [Match] section containing
MACAddress= would match any device without a MAC. This restores the
matching logic prior to e90d037.

(cherry picked from commit 25ea58d37385af27301b7ad25e985eb15f421614)

Gbp-Pq: Name network-link-Fix-logic-error-in-matching-devices-by-MAC.patch

7 years agotest: Drop SKIP_INITRD for QEMU-based tests
Michael Biebl [Mon, 16 Jul 2018 09:27:44 +0000 (11:27 +0200)]
test: Drop SKIP_INITRD for QEMU-based tests

Not all distros support booting without an initrd. E.g. the Debian
kernel builds ext4 as a module and so relies on an initrd to
successfully start the QEMU-based images.

(cherry picked from commit c2d4da002095fe6f86f89a508a81e48fb6d3196f)

Gbp-Pq: Name test-Drop-SKIP_INITRD-for-QEMU-based-tests.patch

7 years agosysusers,tmpfiles: re-create systemd-network, systemd-resolve and systemd-timesync
Yu Watanabe [Sun, 15 Jul 2018 13:31:37 +0000 (22:31 +0900)]
sysusers,tmpfiles: re-create systemd-network, systemd-resolve and systemd-timesync

This partially reverts d4e9e574ea0b5d23598a317e68399584d229568b,
0187368cadea183e18c6d575a9d6b7f491a402af, and
4240cb02fda90ba11dfc0114201e42691132c6a9.
The services systemd-networkd, systemd-resolved, and systemd-timesyncd
enable DynamicUsers= and have bus interfaces. Unfortunately, these
has many problems now. Let us create the relevant users, at least,
tentatively.

Fixes #9503.

(cherry picked from commit 5b5d82615011b9827466b7cd5756da35627a1608)

Gbp-Pq: Name sysusers-tmpfiles-re-create-systemd-network-systemd-resol.patch

7 years agotest: fix networkd-test.py rate limiting and dynamic user
Martin Pitt [Mon, 2 Jul 2018 20:26:31 +0000 (22:26 +0200)]
test: fix networkd-test.py rate limiting and dynamic user

 - Reset systemd-networkd.service before each test run, to avoid running
   into restart limits.
 - Our networkd-test-router.service unit needs to run as root and thus
   can't use `User=`; but networkd still insists on the
   `systemd-network` system user to exist, so create it.

(cherry picked from commit c44c1b8ab5274c7cdb93d9a80c79bcddb503c235)

Gbp-Pq: Name test-fix-networkd-test.py-rate-limiting-and-dynamic-user.patch

7 years agotmpfiles: specify access mode for /run/systemd/netif
Yu Watanabe [Mon, 25 Jun 2018 05:56:49 +0000 (14:56 +0900)]
tmpfiles: specify access mode for /run/systemd/netif

This partially reverts 2af767729489f6baa98a2641b2007acab44ed353.
As the directories are certainly readable and not-writable by
non-privileged users.

(cherry picked from commit 400d8461381c5748a97fccb8a022a20f40063a7d)

Gbp-Pq: Name tmpfiles-specify-access-mode-for-run-systemd-netif.patch

7 years agobus-util: make log level lower in request_name_destroy_callback()
Yu Watanabe [Wed, 27 Jun 2018 12:20:55 +0000 (21:20 +0900)]
bus-util: make log level lower in request_name_destroy_callback()

Fixes #9442.

(cherry picked from commit 04e075b572b168a183cde29480dbb30d027d4fe7)

Gbp-Pq: Name bus-util-make-log-level-lower-in-request_name_destroy_cal.patch

7 years agobasic: add missing comma in raw_clone assembly for sparc
Mike Gilbert [Wed, 27 Jun 2018 14:36:50 +0000 (10:36 -0400)]
basic: add missing comma in raw_clone assembly for sparc

Fixes: 96f64eb5741b157f26a9089816fdd992e959792e
Closes: https://github.com/systemd/systemd/issues/9444
(cherry picked from commit 358248caa3eca525751694de7c34b16bb46d5a9c)

Gbp-Pq: Name basic-add-missing-comma-in-raw_clone-assembly-for-sparc.patch

7 years agosleep: fix one more printf format of a fiemap field
Filipe Brandenburger [Wed, 27 Jun 2018 16:19:33 +0000 (09:19 -0700)]
sleep: fix one more printf format of a fiemap field

Use PRIu64 constant to get the format right on LP-64 architectures,
cast to (uint64_t) to solve incompatibility of __u64.

This was missed in ad4bc3352285f467f4ffa03c3171b19fa0a8758d, so fix it
with this follow up.

(cherry picked from commit d7af62d52cd3b58c5fe7d410e907e01fed71c02e)

Gbp-Pq: Name sleep-fix-one-more-printf-format-of-a-fiemap-field.patch

7 years agotimesync: changes type of drift_freq to int64_t
Yu Watanabe [Sat, 23 Jun 2018 00:41:55 +0000 (09:41 +0900)]
timesync: changes type of drift_freq to int64_t

drift_freq is used for storing timex.freq, and is a 64bit integer.
To support x32 ABI, this changes the type of drift_freq to int64_t.

Fixes #9387.

(cherry picked from commit 75ca1621db4647a4d62d7873cd6715e28fe0f9fa)

Gbp-Pq: Name timesync-changes-type-of-drift_freq-to-int64_t.patch

7 years agosleep: fix printf format of fiemap fields
Filipe Brandenburger [Tue, 26 Jun 2018 16:43:49 +0000 (09:43 -0700)]
sleep: fix printf format of fiemap fields

Use PRIu64 and PRIu32 constants to also get the format right on LP-64
architectures.

For the 64-bit fields, we need a cast to (uint64_t), since __u64 is
defined as a `long long unsigned` and PRIu64 expects a `long unsigned`.
In practice, both are the same, so the cast should be OK.

(cherry picked from commit ad4bc3352285f467f4ffa03c3171b19fa0a8758d)

Gbp-Pq: Name sleep-fix-printf-format-of-fiemap-fields.patch

7 years agosystemd (239-11) unstable; urgency=high
Michael Biebl [Sun, 28 Oct 2018 12:02:18 +0000 (12:02 +0000)]
systemd (239-11) unstable; urgency=high

  [ Michael Biebl ]
  * debian/tests/upstream: Clean up after each test run.
    Otherwise the loopback images used by qemu are not properly released and
    we might run out of disk space.
  * dhcp6: Make sure we have enough space for the DHCP6 option header.
    Fixes out-of-bounds heap write in systemd-networkd dhcpv6 option
    handling.
    (CVE-2018-15688, LP: #1795921, Closes: #912008)
  * chown-recursive: Rework the recursive logic to use O_PATH.
    Fixes a race condition in chown_one() which allows an attacker to cause
    systemd to set arbitrary permissions on arbitrary files.
    (CVE-2018-15687, LP: #1796692, Closes: #912007)

  [ Martin Pitt ]
  * debian/tests/boot-and-services: Use gdm instead of lightdm.
    This seems to work more reliably, on Ubuntu CI's i386 instances lightdm
    fails.

  [ Manuel A. Fernandez Montecelo ]
  * Run "meson test" instead of "ninja test"
    Upstream developers of meson recommend to run it in this way, because
    "ninja test" just calls "meson test", and by using meson directly and
    using extra command line arguments it is possible to control aspects of
    how the tests are run.
  * Increase timeout for test in riscv64.
    The buildds for the riscv64 arch used at the moment are slow, so increase
    the timeouts for this arch by a factor of 10, for good measure.
    (Closes: #906429)

[dgit import unpatched systemd 239-11]

7 years agoImport systemd_239-11.debian.tar.xz
Michael Biebl [Sun, 28 Oct 2018 12:02:18 +0000 (12:02 +0000)]
Import systemd_239-11.debian.tar.xz

[dgit import tarball systemd 239-11 systemd_239-11.debian.tar.xz]

7 years agoImport systemd_239.orig.tar.gz
Michael Biebl [Fri, 22 Jun 2018 22:18:08 +0000 (23:18 +0100)]
Import systemd_239.orig.tar.gz

[dgit import orig systemd_239.orig.tar.gz]