]> git.ipfire.org Git - thirdparty/systemd.git/log
thirdparty/systemd.git
3 years agoci: re-enable s390x Packit builds 22499/head
Frantisek Sumsal [Sun, 13 Feb 2022 14:00:22 +0000 (15:00 +0100)] 
ci: re-enable s390x Packit builds

The s390x Copr builders are finally[0] 'native' instead of QEMU-emulated,
which makes them pretty quick to be used in PRs.

[0] https://lists.fedoraproject.org/archives/list/copr-devel@lists.fedorahosted.org/message/AR3ZDKET3EXZHV3MSU3UHMO7EIKBGAN2/

3 years agoci: explicitly disable multi-status for Super-Linter
Frantisek Sumsal [Sun, 13 Feb 2022 13:59:08 +0000 (14:59 +0100)] 
ci: explicitly disable multi-status for Super-Linter

to, hopefully, get rid of the following error:

```
 2022-02-13 13:32:12 [ERROR]   Failed to get [GITHUB_TOKEN]!
 2022-02-13 13:32:12 [ERROR]   []
 2022-02-13 13:32:12 [ERROR]   Please set a [GITHUB_TOKEN] from the main workflow environment to take advantage of multiple status reports!
```

3 years agoci: use the 'slim' version of Super-Linter
Frantisek Sumsal [Sun, 13 Feb 2022 13:54:51 +0000 (14:54 +0100)] 
ci: use the 'slim' version of Super-Linter

The 'slim' version drops certain storage-heavy linters[0] which we don't
use anyway, so let's make the job a bit faster by downloading and using
a smaller image.

[0] https://github.com/github/super-linter#slim-image

3 years agoboot: Correctly check the return value of CheckEvent
Jan Janssen [Fri, 11 Feb 2022 12:17:23 +0000 (13:17 +0100)] 
boot: Correctly check the return value of CheckEvent

Fixes: #22428
3 years agoMerge pull request #22485 from poettering/cgroup-id-not-support
Luca Boccassi [Fri, 11 Feb 2022 17:06:30 +0000 (17:06 +0000)] 
Merge pull request #22485 from poettering/cgroup-id-not-support

cgroup: downgrade warnings of "cgroup ID" can't be read (on old kernels)

3 years agoci: run all fuzz targets on CIFuzz
Evgeny Vereshchagin [Fri, 11 Feb 2022 02:01:33 +0000 (02:01 +0000)] 
ci: run all fuzz targets on CIFuzz

CIFuzz has been kind of broken for a couple months because
coverage reports downloaded from OSS-Fuzz contain absolute
paths while paths to files changed in PRs are relative and they
don't match. It makes it kind of hard for CIFuzz to figure out
what it should run so it runs either all fuzz targets or just new
fuzz targets. Until that issue is fixed let's just always predictably run
all fuzz targets.

3 years agopid1: export cgroup ID among per-unit cgroup information 22485/head
Lennart Poettering [Fri, 11 Feb 2022 12:24:35 +0000 (13:24 +0100)] 
pid1: export cgroup ID among per-unit cgroup information

It's really interesting for debugging purposes and we have it already,
hence expose it as dbus property.

3 years agocgroup-util: minor modernizations
Lennart Poettering [Fri, 11 Feb 2022 12:23:32 +0000 (13:23 +0100)] 
cgroup-util: minor modernizations

Rename return parameters to "ret", use ternary op without second
argument, rebreak comments, use isempty() more.

3 years agocgroup: downgrade warning if we can't get ID off cgroup
Lennart Poettering [Fri, 11 Feb 2022 12:18:58 +0000 (13:18 +0100)] 
cgroup: downgrade warning if we can't get ID off cgroup

The cgroupid feature was not available in old cgroupvs2 kernels, hence
try to get it but if we can't because it's not supported, then only
debug log about it and proceed.

(We only needs this for cgroup bpf stuff, but that isn't available on
such old kernels anyway)

Fixes: #22483
3 years agobasic: do not warn in mkdir_p() when parent directory exists
Zbigniew Jędrzejewski-Szmek [Thu, 10 Feb 2022 07:30:08 +0000 (08:30 +0100)] 
basic: do not warn in mkdir_p() when parent directory exists

This effectively disables warnings about type/mode/ownership of existing
directories when recursively creating parent directories. (Or files. If there's
a file in a place we expect a directory, the code will later try to create
a file and fail. This follows the general pattern where we do (void)mkdir()
if the mkdir() is immediately followed by opening of a file.)

I was recently debugging an issue with the fstab-generator [1], and it says:
'Directory "/tmp" already exists, but has mode 0777 that is too permissive (0644 was requested), refusing.'
which is very specific but totally wrong in this context.
This output was added in 37c1d5e97dbc869edd8fc178427714e2d9428d2b, and I still
think it is worth to do it, because if you actually *do* want the directory, if
there's something wrong, the precise error message will make it much easier to
diagnose. And we can't easily pass the information what failed up the call chain
because there are multiple things we check (ownership, permission mask, type)…
So passing a param whether to warn or not down into the library code seems like
the best solution, despite not being very elegant.

[1] https://bugzilla.redhat.com/show_bug.cgi?id=2051285

3 years agoresolve: describe mdns event sources
Evgeny Vereshchagin [Thu, 10 Feb 2022 19:12:28 +0000 (19:12 +0000)] 
resolve: describe mdns event sources

when they go down resolved prints
```
Event source mdns-ipv4 (type io) returned error, disabling
```
instead of
```
Event source n/a (type io) returned error, disabling
```

3 years agotree-wide: move `unsigned` to the start of type declaration
Frantisek Sumsal [Thu, 10 Feb 2022 16:19:27 +0000 (17:19 +0100)] 
tree-wide: move `unsigned` to the start of type declaration

Even though ISO C11 doesn't mandate in which order the type specifiers
should appear, having `unsigned` at the beginning of each type
declaration feels more natural and, more importantly, it unbreaks
Coccinelle, which has a hard time parsing `long unsigned` and others:

```
init_defs_builtins: /usr/lib64/coccinelle/standard.h
init_defs: /home/mrc0mmand/repos/systemd/coccinelle/macros.h
HANDLING: src/shared/mount-util.c
: 1: strange type1, maybe because of weird order: long unsigned
```

Most of the codebase already "complies", so let's fix the remaining
"offenders".

3 years agotest: document how to manually run Ubuntu CI stuff
Frantisek Sumsal [Thu, 10 Feb 2022 10:59:27 +0000 (11:59 +0100)] 
test: document how to manually run Ubuntu CI stuff

Every time I need it I have to first relearn autopkgtest and find where
all the necessary stuff lives, so let's document it somewhere close to
systemd for (at least) future me.

3 years agotest: accept GC'ed units in newer LVM
Frantisek Sumsal [Thu, 10 Feb 2022 11:29:53 +0000 (12:29 +0100)] 
test: accept GC'ed units in newer LVM

Since lvm 2.03.15 the transient units are started without `-r`, thus
disappearing once they finish and breaking the test (which expects them
to remain loaded after finishing). Let's accept `LoadState=not-found` as
a valid result as well to fix this.

Follow-up to: d10d562bd4b9f93130fb2b23f2b0d0d4126ea7d4
See: https://sourceware.org/git/?p=lvm2.git;a=commit;h=fbd8b0cf43dc67f51f86f060dce748f446985855

3 years agoMerge pull request #22471 from yuwata/network-bridge-vlan-protocol
Yu Watanabe [Thu, 10 Feb 2022 11:40:04 +0000 (20:40 +0900)] 
Merge pull request #22471 from yuwata/network-bridge-vlan-protocol

network: bridge: fix vlan protocol

3 years agoMerge pull request #22452 from yuwata/network-ipv6ll
Yu Watanabe [Thu, 10 Feb 2022 11:38:56 +0000 (20:38 +0900)] 
Merge pull request #22452 from yuwata/network-ipv6ll

network: use sysctl to set IPv6LL address generation mode when the interface is already up

3 years agotest-network: add missing tests for bridge properties 22471/head
Yu Watanabe [Thu, 10 Feb 2022 09:04:34 +0000 (18:04 +0900)] 
test-network: add missing tests for bridge properties

3 years agonetwork: bridge: fix endian of vlan protocol
Yu Watanabe [Thu, 10 Feb 2022 08:47:14 +0000 (17:47 +0900)] 
network: bridge: fix endian of vlan protocol

Fixes #22469.

3 years agotest-network: add testcase for re-generating IPv6LL address 22452/head
Yu Watanabe [Tue, 8 Feb 2022 14:53:43 +0000 (23:53 +0900)] 
test-network: add testcase for re-generating IPv6LL address

3 years agonetwork: attempt to trigger kernel IPv6LL address generation
Joerie de Gram [Tue, 8 Feb 2022 13:56:26 +0000 (22:56 +0900)] 
network: attempt to trigger kernel IPv6LL address generation

Try to ensure kernel IPv6 link local address generation occurs by
setting the per-if addr_gen_mode sysctl when the link is already up,
instead of the netlink interface (IFLA_INET6_ADDR_GEN_MODE).

The netlink setting is sufficient in cases where the interface is not
yet up when networkd configures an interface - bringing the interface
up will trigger in-kernel address generation.

If the interface is already up, yet the interface has no IPv6LL assigned
setting IFLA_INET6_ADDR_GEN_MODE has no effect.

Writing the addr_gen_mode sysctl is a best effort attempt at triggering
address generation regardless of interface state because it also works
in cases where the interface is already up.

Fixes #22424.

3 years agonetwork: move link_set_ipv6ll_stable_secret() to networkd-ipv6ll.c
Yu Watanabe [Tue, 8 Feb 2022 13:53:33 +0000 (22:53 +0900)] 
network: move link_set_ipv6ll_stable_secret() to networkd-ipv6ll.c

3 years agonetwork: skip to set IPv6LL address generation mode if the requested mode is already set
Yu Watanabe [Tue, 8 Feb 2022 13:27:17 +0000 (22:27 +0900)] 
network: skip to set IPv6LL address generation mode if the requested mode is already set

3 years agonetwork: use log_link_warning_errno() or friends where applicable
Yu Watanabe [Tue, 8 Feb 2022 17:14:12 +0000 (02:14 +0900)] 
network: use log_link_warning_errno() or friends where applicable

3 years agonetwork: monitor current IPv6LL address generation mode
Yu Watanabe [Tue, 8 Feb 2022 13:26:23 +0000 (22:26 +0900)] 
network: monitor current IPv6LL address generation mode

3 years agonetwork: introduce two helper functions for setting IPv6LL address generation mode
Yu Watanabe [Tue, 8 Feb 2022 13:16:38 +0000 (22:16 +0900)] 
network: introduce two helper functions for setting IPv6LL address generation mode

3 years agoudev/net: support Match.Firmware= in .link files (#22462)
Alvin Šipraga [Thu, 10 Feb 2022 07:19:28 +0000 (08:19 +0100)] 
udev/net: support Match.Firmware= in .link files (#22462)

In cbcdcaaa0ec5 ("Add support for conditions on the machines firmware")
a new Firmware= directive was added for .netdev and .network files.
While it was also documented to work on .link files, in actual fact the
support was missing. Add that one extra line to make it work, and also
update the fuzzer directives.

3 years agoMerge pull request #22467 from mrc0mmand/TEST-64-lvm-static-autoactivation
Anita Zhang [Thu, 10 Feb 2022 01:10:06 +0000 (17:10 -0800)] 
Merge pull request #22467 from mrc0mmand/TEST-64-lvm-static-autoactivation

test: lvm 2.03.15 dropped the static autoactivation

3 years agotest: lvm 2.03.15 dropped the static autoactivation 22467/head
Frantisek Sumsal [Wed, 9 Feb 2022 21:35:03 +0000 (22:35 +0100)] 
test: lvm 2.03.15 dropped the static autoactivation

so install the respective generator only if we're running with older
lvm versions.

See: https://sourceware.org/git/?p=lvm2.git;a=commit;h=ee8fb0310c53ed003a43b324c99cdfd891dd1a7c

3 years agoClarify protocol used in systemd-journal-upload (#22465)
Federico Ceratto [Wed, 9 Feb 2022 20:34:29 +0000 (20:34 +0000)] 
Clarify protocol used in systemd-journal-upload (#22465)

* Clarify protocol used in systemd-journal-upload

3 years agoMerge pull request #22458 from poettering/parse-safe-string
Luca Boccassi [Wed, 9 Feb 2022 16:59:14 +0000 (16:59 +0000)] 
Merge pull request #22458 from poettering/parse-safe-string

some safety tweaks to conf-parser.[ch]

3 years agomeson: disable export-dbus-interfaces target when cross-compiling
Luca Boccassi [Tue, 8 Feb 2022 13:19:52 +0000 (13:19 +0000)] 
meson: disable export-dbus-interfaces target when cross-compiling

ERROR:
Cannot use target systemd as a generator because it is built for the
host machine and no exe wrapper is defined or needs_exe_wrapper is
true. You might want to set `native: true` instead to build it for
the build machine.

3 years agoci: remove MULTI_STATUS from superlinter
Evgeny Vereshchagin [Tue, 8 Feb 2022 20:35:17 +0000 (20:35 +0000)] 
ci: remove MULTI_STATUS from superlinter

Judging by
ERROR! Failed to call GitHub Status API!

it doesn't seem to work. Even if it did it would just clutter the status
checks I think so let's just remove MULTI_STATUS along with
GITHUB_TOKEN.

3 years agoMerge pull request #22248 from yuwata/network-tc-use-request-queue
Yu Watanabe [Wed, 9 Feb 2022 09:57:37 +0000 (18:57 +0900)] 
Merge pull request #22248 from yuwata/network-tc-use-request-queue

network: tc: use request queue to configure QDisc or TClass

3 years agoconf-parser: add specific parser for PID values 22458/head
Lennart Poettering [Wed, 9 Feb 2022 09:13:59 +0000 (10:13 +0100)] 
conf-parser: add specific parser for PID values

3 years agotree-wide: use config_parse_safe_string() at various places
Lennart Poettering [Wed, 9 Feb 2022 09:07:17 +0000 (10:07 +0100)] 
tree-wide: use config_parse_safe_string() at various places

3 years agoconf-parser: update config_item_*_lookup() to follow modern coding style
Lennart Poettering [Wed, 9 Feb 2022 08:55:21 +0000 (09:55 +0100)] 
conf-parser: update config_item_*_lookup() to follow modern coding style

Let's rename the return parameters ret_xyz, and always initialize them
if we return >= 0, as per our current coding style.

3 years agoconf-parser: refuse section headers with control characters early, just in case
Lennart Poettering [Wed, 9 Feb 2022 08:47:53 +0000 (09:47 +0100)] 
conf-parser: refuse section headers with control characters early, just in case

3 years agoconf-parser: minor refactorings/modernizations
Lennart Poettering [Wed, 9 Feb 2022 08:47:50 +0000 (09:47 +0100)] 
conf-parser: minor refactorings/modernizations

3 years agoconf-parser: use _cleanup_free_ where appropriate
Lennart Poettering [Wed, 9 Feb 2022 08:47:10 +0000 (09:47 +0100)] 
conf-parser: use _cleanup_free_ where appropriate

3 years agoconf-parse: add generic config_parse_safe_string() helper
Lennart Poettering [Wed, 9 Feb 2022 08:45:12 +0000 (09:45 +0100)] 
conf-parse: add generic config_parse_safe_string() helper

This helper is just like config_parse_string() but does some superficial
checks for control characters and quotes.

In most cases we currently use config_parse_string() we probably want to
use config_parse_safe_string() for safety reasons.

3 years agoAdd support for `isolated` parameter
Santa Wiryaman [Mon, 3 May 2021 22:48:26 +0000 (18:48 -0400)] 
Add support for `isolated` parameter

Add the "Isolated" parameter in the *.network file, e.g.,

[Bridge]
Isolated=true|false

When the Isolated parameter is true, traffic coming out of this port
will only be forward to other ports whose Isolated parameter is false.

When Isolated is not specified, the port uses the kernel default
setting (false).

The "Isolated" parameter was introduced in Linux 4.19.
See man bridge(8) for more details.
But even though the kernel and bridge/iproute2 recognize the "Isolated"
parameter, systemd-networkd did not have a way to set it.

3 years agoci: validate actions and fix actionlint warnings
Evgeny Vereshchagin [Tue, 8 Feb 2022 17:34:10 +0000 (17:34 +0000)] 
ci: validate actions and fix actionlint warnings

some actions like Coverity and CFLite aren't run on every PR so to make
sure they are more or less fine when they are changed it makes sense to
at least check them with superlinter/actionlint: https://github.com/rhysd/actionlint

The following warnings were fixed along the way:
```
.github/workflows/mkosi.yml:55:7: shellcheck reported issue in this script: SC2086:info:6:14: Double quote to prevent globbing and word splitting [shellcheck]
   |
55 |       run: |
   |       ^~~~
.github/workflows/mkosi.yml:55:7: shellcheck reported issue in this script: SC2046:warning:6:40: Quote this to prevent word splitting [shellcheck]
   |
55 |       run: |
   |       ^~~~
.github/workflows/mkosi.yml:55:7: shellcheck reported issue in this script: SC2006:style:6:40: Use $(...) notation instead of legacy backticked `...` [shellcheck]
   |
55 |       run: |
   |       ^~~~
```
```
.github/workflows/coverity.yml:31:9: shellcheck reported issue in this script: SC2086:info:1:93: Double quote to prevent globbing and word splitting [shellcheck]
   |
31 |         run: echo "COVERITY_SCAN_NOTIFICATION_EMAIL=$(git log -1 ${{ github.sha }} --pretty=\"%aE\")" >> $GITHUB_ENV
   |         ^~~~
```

3 years agonetwork: tc: use hashmap to store traffic control settings in .network files 22248/head
Yu Watanabe [Wed, 26 Jan 2022 04:06:37 +0000 (13:06 +0900)] 
network: tc: use hashmap to store traffic control settings in .network files

As now each qdisc or tclass is configured when it is ready.

3 years agonetwork: tc: introduce order dependency of traffic control
Yu Watanabe [Wed, 26 Jan 2022 04:01:19 +0000 (13:01 +0900)] 
network: tc: introduce order dependency of traffic control

3 years agonetwork: tc: use request queue to configure traffic control
Yu Watanabe [Tue, 25 Jan 2022 18:49:27 +0000 (03:49 +0900)] 
network: tc: use request queue to configure traffic control

But no dependency resolution is implemented.

3 years agonetwork: tc/teql: set tca_kind in verify()
Yu Watanabe [Wed, 26 Jan 2022 03:18:47 +0000 (12:18 +0900)] 
network: tc/teql: set tca_kind in verify()

Then, qdisc_configure() can be simplified.

3 years agonetwork: tc: monitor qdisc and tclass
Yu Watanabe [Tue, 25 Jan 2022 23:42:22 +0000 (08:42 +0900)] 
network: tc: monitor qdisc and tclass

3 years agosd-netlink: allow to dump qdisc and tclass
Yu Watanabe [Wed, 26 Jan 2022 01:24:50 +0000 (10:24 +0900)] 
sd-netlink: allow to dump qdisc and tclass

3 years agosd-netlink: make traffic control related message can be monitored
Yu Watanabe [Wed, 26 Jan 2022 00:53:40 +0000 (09:53 +0900)] 
sd-netlink: make traffic control related message can be monitored

3 years agosd-netlink: introduce sd_rtnl_message_traffic_control_get_{ifindex,parent,handle}()
Yu Watanabe [Tue, 25 Jan 2022 21:04:16 +0000 (06:04 +0900)] 
sd-netlink: introduce sd_rtnl_message_traffic_control_get_{ifindex,parent,handle}()

3 years agosd-netlink: unify sd_rtnl_message_new_qdisc() and sd_rtnl_message_new_tclass()
Yu Watanabe [Tue, 25 Jan 2022 21:00:55 +0000 (06:00 +0900)] 
sd-netlink: unify sd_rtnl_message_new_qdisc() and sd_rtnl_message_new_tclass()

As the netlink message header types for both qdisc and tclass are
equivalent.

3 years agonetwork: coding style fixes
Yu Watanabe [Wed, 26 Jan 2022 05:28:17 +0000 (14:28 +0900)] 
network: coding style fixes

3 years agonetwork: tc: drop unnecessary conditions
Yu Watanabe [Tue, 25 Jan 2022 21:42:23 +0000 (06:42 +0900)] 
network: tc: drop unnecessary conditions

As TC_H_UNSPEC is 0.

3 years agonetwork: tc: drop unused QDisc::family element
Yu Watanabe [Tue, 25 Jan 2022 20:52:54 +0000 (05:52 +0900)] 
network: tc: drop unused QDisc::family element

3 years agoUpdate footer.html
Eduard Tolosa [Thu, 3 Feb 2022 18:25:57 +0000 (13:25 -0500)] 
Update footer.html

3 years agoMerge pull request #22446 from poettering/sd-boot-cpuid-vm-tscd
Lennart Poettering [Tue, 8 Feb 2022 20:57:38 +0000 (21:57 +0100)] 
Merge pull request #22446 from poettering/sd-boot-cpuid-vm-tscd

sd-boot: don't read TSC in VM environments, and other TSC fixes

3 years agonetwork: move ipv6ll related functions to networkd-ipv6ll.[ch]
Yu Watanabe [Tue, 8 Feb 2022 12:56:02 +0000 (21:56 +0900)] 
network: move ipv6ll related functions to networkd-ipv6ll.[ch]

3 years agoMerge pull request #22448 from poettering/coredump-raise-sizes
Yu Watanabe [Tue, 8 Feb 2022 16:23:07 +0000 (01:23 +0900)] 
Merge pull request #22448 from poettering/coredump-raise-sizes

coredump: raise limits

3 years agoRevert "ci: switch to fedora-35 on i386 on Packit"
Yu Watanabe [Sat, 22 Jan 2022 18:07:52 +0000 (03:07 +0900)] 
Revert "ci: switch to fedora-35 on i386 on Packit"

This reverts commit 478c632e00ffbf50012b1946a8a025494c012353.

3 years agoRevert "packit: switch the remaining jobs to F35"
Yu Watanabe [Sat, 22 Jan 2022 18:07:18 +0000 (03:07 +0900)] 
Revert "packit: switch the remaining jobs to F35"

This reverts commit e7642152ae6e867fc569206e905464321a41e58a.

3 years agoudev/net: support to set MDI-X mode
Yu Watanabe [Fri, 4 Feb 2022 04:06:27 +0000 (13:06 +0900)] 
udev/net: support to set MDI-X mode

Closes #22386.

3 years agoMerge pull request #22423 from poettering/repart-sector-size
Lennart Poettering [Tue, 8 Feb 2022 13:47:00 +0000 (14:47 +0100)] 
Merge pull request #22423 from poettering/repart-sector-size

repart: fix sector size handling

3 years agowatchdog: saturate to kernel's max watchdog timeout
Curtis Klein [Sun, 10 Oct 2021 01:18:54 +0000 (18:18 -0700)] 
watchdog: saturate to kernel's max watchdog timeout

Since version 4.5, the max possible timeout is UINT_MAX / 1000 since it
does calculations in milliseconds. A small helper function is added to
make this conversion and saturation and will be used more in the next
commit.

Also document the usage of signed integers by the kernel userspace API.

3 years agoMerge pull request #22442 from DaanDeMeyer/fix-entry-iteration
Lennart Poettering [Tue, 8 Feb 2022 13:29:24 +0000 (14:29 +0100)] 
Merge pull request #22442 from DaanDeMeyer/fix-entry-iteration

journal: Fix entry array iteration in case of corruption

3 years agoMerge pull request #22366 from poettering/journal-file-punch-fix
Daan De Meyer [Tue, 8 Feb 2022 13:03:03 +0000 (13:03 +0000)] 
Merge pull request #22366 from poettering/journal-file-punch-fix

journald: some journal file hole punching fixes

3 years agosd-boot: don't use TSC in virtualized environments 22446/head
Lennart Poettering [Tue, 8 Feb 2022 10:35:10 +0000 (11:35 +0100)] 
sd-boot: don't use TSC in virtualized environments

Replaces: f699bd81e8e18da2d2fc11e7fb7dce95f8bb3f9e

Fixes: #22060
3 years agosd-boot: return 0 (not 1) from ticks_read() in fallback implementation
Lennart Poettering [Tue, 8 Feb 2022 10:27:48 +0000 (11:27 +0100)] 
sd-boot: return 0 (not 1) from ticks_read() in fallback implementation

The single consumer of ticks_read() (i.e. time_usec()) checks for == 0
to detect the "not supported/invalid" case, hence actually return the
right value for that.

3 years agosd-boot: encapsulate freq cache in ticks_freq()
Lennart Poettering [Tue, 8 Feb 2022 10:26:49 +0000 (11:26 +0100)] 
sd-boot: encapsulate freq cache in ticks_freq()

It's usually nicer to abstract a cache away in the provider of a value,
hence do so.

3 years agosd-boot: don't export ticks_read() and ticks_freq()
Lennart Poettering [Tue, 8 Feb 2022 10:25:07 +0000 (11:25 +0100)] 
sd-boot: don't export ticks_read() and ticks_freq()

They only have a single user in time_usec(), hence don't expose them.

3 years agosd-boot: add overflow check to TSC reads
Lennart Poettering [Tue, 8 Feb 2022 10:23:19 +0000 (11:23 +0100)] 
sd-boot: add overflow check to TSC reads

On some archs the counter is 32bit, and in case of virtualization it
might actually overflow, who knows.

3 years agosd-boot: split out TSC/time API
Lennart Poettering [Tue, 8 Feb 2022 10:18:44 +0000 (11:18 +0100)] 
sd-boot: split out TSC/time API

These are a whole family of functions, let's give them their own .c/.h
file.

No code changes, just splitting things out.

3 years agoRevert "boot-timestamps: Discard firmware init time when running in a VM"
Lennart Poettering [Thu, 3 Feb 2022 11:06:54 +0000 (12:06 +0100)] 
Revert "boot-timestamps: Discard firmware init time when running in a VM"

This reverts commit f699bd81e8e18da2d2fc11e7fb7dce95f8bb3f9e (#22063)

We should suppress the TSC data when we generate it if we assume its
invalid, not when we consume it, because at that point we don't even
know if the data stems from TSC or something else.

3 years agodocs: Improve HOME_DIRECTORY.md
adrian5 [Tue, 8 Feb 2022 11:42:19 +0000 (12:42 +0100)] 
docs: Improve HOME_DIRECTORY.md

3 years agoMerge pull request #22430 from yuwata/resolve-fix-memleak-and-use-after-free
Luca Boccassi [Tue, 8 Feb 2022 12:01:14 +0000 (12:01 +0000)] 
Merge pull request #22430 from yuwata/resolve-fix-memleak-and-use-after-free

resolve: fix memleak and use-after-free, and several cleanups

3 years agoman: coredump.conf: document defaults limits 22448/head
Lennart Poettering [Tue, 8 Feb 2022 10:54:37 +0000 (11:54 +0100)] 
man: coredump.conf: document defaults limits

3 years agocoredump: raise the coredump save size on 64bit systems to 32G (and lower it to 1G...
Lennart Poettering [Tue, 8 Feb 2022 10:52:17 +0000 (11:52 +0100)] 
coredump: raise the coredump save size on 64bit systems to 32G (and lower it to 1G on 32bit systems)

Apparently 2G is too low for various real-life systems. But raising it
universally above 2^32 sounds wrong to me, since that makes no sense on
32bit systems, that we still support.

Hence, let's raise the limit to 32G on 64bit systems, and *lower* it to
1G on 32bit systems.

32G is 4 orders of magnitude higher then the old settings. Let's hope
that's enough for now. Should this not be enough we can raise it
further.

Fixes: #22076
3 years agoREADME: CONFIG_KCMP is the new kernel compile time option for kcmp()
Lennart Poettering [Tue, 8 Feb 2022 10:45:00 +0000 (11:45 +0100)] 
README: CONFIG_KCMP is the new kernel compile time option for kcmp()

Fixes: #21947
3 years agotests: add repart tests for block devices with 1024, 2048, 4096 byte sector sizes 22423/head
Lennart Poettering [Mon, 7 Feb 2022 13:35:40 +0000 (14:35 +0100)] 
tests: add repart tests for block devices with 1024, 2048, 4096 byte sector sizes

let's make sure repart works with 4K drives and exotic sector sizes.

3 years agorepart: fix sector size handling
Lennart Poettering [Fri, 4 Feb 2022 16:39:44 +0000 (17:39 +0100)] 
repart: fix sector size handling

This queries the sector size from libfdisk instead of assuming 512, and
uses that when converting from bytes to the offset/size values libfdisk
expects.

This is an alternative to Tom Yan's #21823, but prefers using libfdisk's
own ideas of the sector size instead of going directly to the backing
device via ioctls. (libfdisk can after all also operate on regular
files, where the sector size concept doesn't necessarily apply the same
way.)

This also makes the "grain" variable, i.e. how we'll align the
partitions. Previously this was hardcoded to 4K, and that still will be
the minimum grain we use, but should the sector size be larger than that
we'll use the next multiple of the sector size instead.

3 years agoresolve: reuse timer event source for DnsQuery 22430/head
Yu Watanabe [Sat, 5 Feb 2022 13:31:06 +0000 (22:31 +0900)] 
resolve: reuse timer event source for DnsQuery

If the query get CNAME or DNAME, then the query will be restarted.
Even in that case, previously, the event source was freed and allocated
again. Let's slightly optimize it.

3 years agoresolve: use _cleanup_ attribute for freeing DnsQuery
Yu Watanabe [Sat, 5 Feb 2022 13:04:42 +0000 (22:04 +0900)] 
resolve: use _cleanup_ attribute for freeing DnsQuery

3 years agoresolve: fix possible memleak
Yu Watanabe [Sat, 5 Feb 2022 13:03:19 +0000 (22:03 +0900)] 
resolve: fix possible memleak

Fortunately, unlike the issue fixed in the previous commit, the memleak
should be superficial and not become apparent, as the queries handled
here are managed by the stub stream, and will be freed when the stream
is closed.

Just for safety, and slightly reducing the runtime memory usage by the
stub stream.

3 years agoresolve: fix potential memleak and use-after-free
Yu Watanabe [Sat, 5 Feb 2022 12:37:01 +0000 (21:37 +0900)] 
resolve: fix potential memleak and use-after-free

When stub stream is closed early, then queries associated to the stream
are freed. Previously, the timer event source for queries may not be
disabled, hence may be triggered with already freed query.
See also dns_stub_stream_complete().

Note that we usually not set NULL or zero when freeing simple objects.
But, here DnsQuery is large and complicated object, and the element may
be referenced in subsequent freeing process in the future. Hence, for
safety, let's set NULL to the pointer.

3 years agohwbd: 60-sensor.hwdb: Add Pipo W2Pro
Matt Walton [Fri, 4 Feb 2022 17:47:08 +0000 (11:47 -0600)] 
hwbd: 60-sensor.hwdb: Add Pipo W2Pro

Add accelerometer matrix for Pipo W2Pro tablet. Y-axis was backwards (normal <--> bottom-up).

3 years agooom: Cleanup of information dump code after kill
Benjamin Berg [Mon, 7 Feb 2022 16:34:21 +0000 (17:34 +0100)] 
oom: Cleanup of information dump code after kill

This is a follow up to 29f4185a9cdc ("oomd: Dump top offenders after a
kill action") to clean up the code a bit for review comments that
happened after the code had been merged already.

3 years agowatchdog: work around Coverity confusion
Lennart Poettering [Mon, 7 Feb 2022 09:59:05 +0000 (10:59 +0100)] 
watchdog: work around Coverity confusion

Coverity (and I, initially) get really confused about "fn"'s validity
here. it doesn't grok that free_and_strdup() is actually a NOP in case
the string isn't changed, and assumes it always invalidates the
specified buffer, which it doesn't do though.

Follow-up for: f4b1a6b6419b771d26342a9b75b1d77ee9d90133

3 years agojournal: Improve handling of corruption during upwards entry iteration 22442/head
Daan De Meyer [Mon, 7 Feb 2022 20:19:29 +0000 (20:19 +0000)] 
journal: Improve handling of corruption during upwards entry iteration

If we're going upwards in the journal file during entry iteration and we
can't reach the current entry due to corruption, start iterating upwards
from the last reachable entry array. This is equivalent to skipping
all entries in the array that can't be reached anymore.

Fixes #22431

3 years agojournal: Fix upwards iteration of entry items in case of corruption
Daan De Meyer [Mon, 7 Feb 2022 20:15:07 +0000 (20:15 +0000)] 
journal: Fix upwards iteration of entry items in case of corruption

8d801e35cb155faa08235a5af8b4d6ad60715837 didn't take into account
upwards iteration of entry items when we're working on a corrupted
journal file. Instead of moving to the previous entry array, we'd
always move to the next array, regardless of the iteration direction.

To fix this, we introduce bump_entry_array() that moves to the next
or previous entry array depending on the given direction. Since the
entry array chains are singly linked lists, we have to start iterating
from the front to find the previous array. We only reach this logic
if we're working on a corrupted journal file so being slow here shouldn't
matter too much.

3 years agoMerge pull request #22383 from yuwata/network-generator-keep-configuration
Luca Boccassi [Mon, 7 Feb 2022 18:06:24 +0000 (18:06 +0000)] 
Merge pull request #22383 from yuwata/network-generator-keep-configuration

network: enable KeepConfiguration= when running on network filesystem

3 years agologin: typo fix
Michael Biebl [Mon, 7 Feb 2022 16:49:26 +0000 (17:49 +0100)] 
login: typo fix

3 years agoman: fix typo
Michael Biebl [Mon, 7 Feb 2022 13:57:00 +0000 (14:57 +0100)] 
man: fix typo

3 years agoUpdate CODING_STYLE.md
bearhoney [Mon, 7 Feb 2022 10:41:41 +0000 (11:41 +0100)] 
Update CODING_STYLE.md

Fixed typo.

3 years agotests: also fuzz packets sent in the DHCP6_STATE_SOLICITATION state
Evgeny Vereshchagin [Fri, 4 Feb 2022 15:17:25 +0000 (15:17 +0000)] 
tests: also fuzz packets sent in the DHCP6_STATE_SOLICITATION state

With aborts enabled the fuzzer can catch issues like
https://github.com/systemd/systemd/commit/26a63b81322a3bd8b9fbd43f75897c391708de2c
Let's extend it a bit to let it cover issues like
https://github.com/systemd/systemd/pull/22406#discussion_r798932098

3 years agosd-dhcp6-client: fix sending prefix delegation request during rebind
Thomas Haller [Thu, 3 Feb 2022 17:55:18 +0000 (18:55 +0100)] 
sd-dhcp6-client: fix sending prefix delegation request during rebind

Fixes an assertion failure "pd->type == SD_DHCP6_OPTION_IA_PD" in dhcp6_option_append_pd().

Something similar was done in commit 26a63b81322a ('sd-dhcp6-client: Fix
sending prefix delegation request (#17136)'). The justification is
probably the same.

3 years agonetwork: enable KeepConfiguration= when running on network filesystem 22383/head
Yu Watanabe [Fri, 4 Feb 2022 06:33:38 +0000 (15:33 +0900)] 
network: enable KeepConfiguration= when running on network filesystem

Also, set KeepConfiguration=dhcp-on-stop by default when running in
initrd.

Fixes #21967.

3 years agoMerge pull request #22416 from fbuihuu/misc
Yu Watanabe [Sat, 5 Feb 2022 00:26:40 +0000 (09:26 +0900)] 
Merge pull request #22416 from fbuihuu/misc

Misc

3 years agoMerge pull request #22395 from benzea/benzea/oomd-dump-offenders
Anita Zhang [Fri, 4 Feb 2022 22:44:04 +0000 (14:44 -0800)] 
Merge pull request #22395 from benzea/benzea/oomd-dump-offenders

oomd: Dump top offenders after a kill action

3 years agooomd: Dump top offenders after a kill action 22395/head
Benjamin Berg [Thu, 3 Feb 2022 14:50:31 +0000 (15:50 +0100)] 
oomd: Dump top offenders after a kill action

This hopefully makes it more transparent why a specific cgroup was
killed by systemd-oomd.

3 years agoman: systemd-stdio-bridge doesn't connect "two busses"
Lennart Poettering [Thu, 3 Feb 2022 14:57:00 +0000 (15:57 +0100)] 
man: systemd-stdio-bridge doesn't connect "two busses"

In D-Bus, clients connect to a bus (the usual case), or use direct
questions to each other (the unusual case). A bus is a program one can
connect to and implemented by dbus-daemon or dbus-broker. HOwever,
busses never connect between each other, that doesn't exist. Hence don't
claim so.

This is probably confusion about the fact that sd-bus calls D-Bus
connection objects just "sd_bus" for simplicity, given they are used in
99% of the cases to connect to a bus — only in exceptional cases they
are used for direct connections between peers without involving a bus.

Follow-up for b7bb58ef70b0c876941a1c31ed4e2f5f1dc5ed0e

3 years agojournal-file: explicitly handle file systems that do not support hole punching 22366/head
Lennart Poettering [Wed, 2 Feb 2022 09:42:37 +0000 (10:42 +0100)] 
journal-file: explicitly handle file systems that do not support hole punching

Apparently the error code fallocate() returns if hole punching is not
supported is not too well defined (man page just says "an error is
returned"), hence let's accept the usual set of errors, and the
normalize it to EOPNOTSUPP, and generate a clear error message in this
case.