summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2015-10-24util-lib: rework extract_first_word_and_warn() a bitLennart Poettering
- Really warn in all error cases, not just some. We need to make sure that all errors are logged to not confuse the user. - Explicitly check for EINVAL error code before claiming anything about invalid escapes, could be ENOMEM after all.
2015-10-24util: split out extract_first_word() and related calls into extract-word.[ch]Lennart Poettering
This is quite a lot of code these days, hence move it to its own source file.
2015-10-24path-util: unify how we process paths specified on the command lineLennart Poettering
Let's introduce a common function that makes relative paths absolute and warns about any errors while doing so.
2015-10-24util-lib: get_current_dir_name() can return errors other than ENOMEMLennart Poettering
get_current_dir_name() can return a variety of errors, not just ENOMEM, hence don't blindly turn its errors to ENOMEM, but return correct errors in path_make_absolute_cwd(). This trickles down into a couple of other functions, some of which receive unrelated minor fixes too with this commit.
2015-10-24path-util: rework find_binary(), fsck_exists() and mkfs_exists()Lennart Poettering
Modernize the code a bit: - Get rid of FOREACH_WORD_SEPARATOR() loop in favour of a extract_first_word() loop. - Remove find_binary()'s "local" flag. It's not reasonably possible to look for binaries on remote systems, we hence should not pretend we could. - When we cannot find a suitable binary, return the last error returned from access() rather than ENOENT unconditionally. - Rework fsck_exists() and mkfs_exists() to return 1 on success, 0 if the implementation is missing and negative on real errors. This is more like we do it in other functions. - Make sure we also detect direct fsck symlinks to "true", rather than just absolute ones to /bin/true.
2015-10-24Merge pull request #1664 from again4you/devel/tmp-smack_#5Lennart Poettering
mount: add new SmackFileSystemRoot= setting for mount unit (v4)
2015-10-24units: add 'SmackFileSystemRoot=*' option into tmp.mountSangjung Woo
If SMACK is enabled, 'smackfsroot=*' option should be specified when /tmp is mounted since many non-root processes use /tmp for temporary usage. If not, /tmp is labeled as '_' and smack denial occurs when writing. In order to do that, 'SmackFileSystemRoot=*' is newly added into tmp.mount.
2015-10-24mount: add new SmackFileSystemRoot= setting for mount unitSangjung Woo
This option specifies the label to assign the root of the file system if it lacks the Smack extended attribute. Note that this option will be ignored if kernel does not support the Smack feature by runtime checking.
2015-10-24Merge pull request #1657 from phomes/tmpfiles-simplificationLennart Poettering
tmpfiles: simplify mountpoint detection
2015-10-24Merge pull request #1659 from vcaputo/journal_verify_envalidLennart Poettering
journal: s/Envalid/Invalid/
2015-10-23journal: s/Envalid/Invalid/Vito Caputo
2015-10-23tmpfiles: simplify mountpoint detectionThomas Hindoe Paaboel Andersen
No functional change. Just a simplification. A || (!A && B) is the same as A || B Introduced in 78a92a5a2306709e4587e332728a76901323ade9
2015-10-23Merge pull request #1653 from keszybz/lz4-compress-timeLennart Poettering
Limit test-compress-benchmark to approx. 12 s of runtime
2015-10-23test-compress-benchmark: properly initialize bufferZbigniew Jędrzejewski-Szmek
We were compressing unitialized memory, which should not result in any problems, but is inelegant.
2015-10-23test-compress-benchmark: limit default runtime to 2 seconds per subtestZbigniew Jędrzejewski-Szmek
If both lz4 and xz are enabled, this results in a limit of 2×3×2 s ~= 12 s runtime. Previous implementation started with really small buffer sizes. When combined with a short time limit this resulteded in abysmal results for xz. It seems that the initialization overead is really significant for small buffers. Since xz will not be used by default anymore, this does not seem worth fixing. Instead buffer sizes are changed to run a pseudo-random non-repeating pattern. This should allow reasonable testing for all buffer sizes. For testing, both runtime and the buffer size seed can be specified on the command line. Sufficiently large runtime allows all buffer sizes up to 1MB to be tested.
2015-10-23Merge pull request #1652 from filbranden/extract1Lennart Poettering
util: Replace state with separate booleans in extract_first_word
2015-10-23Merge pull request #1647 from evverx/use-extract-in-namespace-parsingLennart Poettering
core: use extract_first_word for namespace parsing
2015-10-23Merge pull request #1648 from evverx/clarify-journalctl-quiet-optionLennart Poettering
journalctl: clarify -q option
2015-10-23Merge pull request #1641 from poettering/btrfs-quotaDaniel Mack
btrfs quota beef up and various other unrelated changes
2015-10-22util: Replace state with separate booleans in extract_first_wordFilipe Brandenburger
This simplifies the logic and uniformizes the way single and double quotes are handled. In the end, the code is about 40 lines shorter. Tested by running the excellent test cases from test-util. Also installed the systemd binaries including this patch and booted a system with it, everything looked normal.
2015-10-22journalctl: clarify -q optionEvgeny Vereshchagin
-q suppresses info messages too
2015-10-22core: use extract_first_word for namespace parsingEvgeny Vereshchagin
see https://github.com/systemd/systemd/pull/1632#issuecomment-149903791 We should port this loop over to extract_first_word(), too.
2015-10-22Merge pull request #1646 from keszybz/man-dnf-installLennart Poettering
dnf install command update
2015-10-22man: also add --enablerepo=updates to dnf invocationZbigniew Jędrzejewski-Szmek
Without the updates repo, we are installing packages from the time that that version of Fedora was released. Normally, during the lifetime of the release most packages are updated, so most of the packages installed would be outdated, and the first update after installation would update a massive set of packages. Avoid all this by installing from the updates repo from the start.
2015-10-22Merge pull request #1625 from teg/ndiscPatrik Flykt
sd-ndisc: Rename (from sd-icmp6-nd) and prepare for handling SLAAC and router discovery
2015-10-22man: remove --nogpg from dnf install command lineZbigniew Jędrzejewski-Szmek
Keys for previous and future Fedora distributions were added for the fedora-repos package recently: https://bugzilla.redhat.com/show_bug.cgi?id=1246701. There is no need to skip signature checking. Also, update to the latest and greatest and remove unnecessary quotes.
2015-10-22sd-ndisc: reduce callbacksTom Gundersen
Only callback on error when the statemachine is in a truly broken state. This is now only the case when we fail to rearm a timer.
2015-10-22sd-ndisc: drop sd_ndisc_get_prefixlen()Tom Gundersen
This is unused, and in the future we will pass prefixes and prefixlengths directly to the callbacks when needed rather than having to search for them.
2015-10-22sd-ndisc: drop sd_ndisc_prefix_match()Tom Gundersen
This is no longer used.
2015-10-22sd-ndisc: don't trigger timeout on prefix expiryTom Gundersen
The caller of the library is no longer notified, so triggering a timer just to clean up is not necessary. Instead check for and clean up invalid prefixes lazily.
2015-10-22sd-ndisc: remove a prefix from the ndisc object when freeing itTom Gundersen
This follows the coding style, and allows some simplification to the rest of the code.
2015-10-22sd-ndisc: don't inform the caller of expired prefixesTom Gundersen
The caller should push any lifetime information into the kernel and let the kernel handle prefix expiration.
2015-10-22sd-ndisc: clean up state enumTom Gundersen
There is no need to assign valuse to the states. Also add _INVALID and _MAX, even though these are not used, it keeps it consistent.
2015-10-22networkd: dhcp6 - do not handle prefix expirationTom Gundersen
This ressurects 47d45d3cde45d6545367570264e4e3636bc9e345. We now always use /128 prefixes, so there is no need for the DHCPv6 code to know about prefixes expiring.
2015-10-22networkd: dhcp6 - DHCPv6 addresses should always be /128Tom Gundersen
The routing information should be configured separately by ND, there is no need to indicate the prefix again in the DHCPv6 addresses. See discussion and related links at issue #1520.
2015-10-22networkd: ndisc - split out from dhcp6 codeTom Gundersen
2015-10-22networkd: ndisc - make logging messages uniformTom Gundersen
Refer to Router Discovery rather than ICMPv6.
2015-10-22sd-ndisc: rename icmp6 to ndisc throughout the codeTom Gundersen
2015-10-22libsystemd-network: split icm6-util.[ch] out of dhcp6 codeTom Gundersen
2015-10-22networkd: rename icmp6 to ndiscTom Gundersen
2015-10-22sd-ndisc: rename API from sd-icmp6-ndTom Gundersen
2015-10-22sd-icmp6-nd: rename files to sd-ndiscTom Gundersen
The actual code rename will follow. The reason for the change of name is to make it simpler and more uniform with how we name other libraries (we don't include the underlying protocol). The new name also matches the naming in the kernel (which is particularly relevent here as we expect to let the kernel do some parts of the protocol and we do others).
2015-10-22Merge pull request #1644 from reverendhomer/patch-1Tom Gundersen
networkd-manager: fix swapped arguments
2015-10-22Merge pull request #1642 from evverx/allow-unbalanced-double-quote-in-relax-modeLennart Poettering
util: allow unbalanced double quote in EXTRACT_QUOTES|EXTRACT_RELAX mode
2015-10-22networkd-manager: fix swapped argumentsreverendhomer
fixes Coverity #1328493
2015-10-22util: allow unbalanced double quote in EXTRACT_QUOTES|EXTRACT_RELAX modeEvgeny Vereshchagin
extract_first_word understands "\'string" but doesn't understand "\"string" fixed this inconsistency.
2015-10-22core: check parsed bus msg in full before applying itLennart Poettering
2015-10-22nspawn: don't try to resolve passed binary before entering namespaceLennart Poettering
Othewise we might follow the symlinks on the host, instead of the container. Fixes #1400
2015-10-22nspawn: rework how we determine private networking settingsLennart Poettering
Make sure we acquire CAP_NET_ADMIN if we require virtual networking. Make sure we imply virtual ethernet correctly when bridge is request. Fixes: #1511 Fixes: #1554 Fixes: #1590
2015-10-22units: also whitelist "blkext" block devices for nspawn serviceLennart Poettering
/dev/loop*p* block devices are of the "blkext" subsystem, not of loop, hence whitelist this too. Fixes #1446