system/systemd
system/systemd
1
0
Fork 0
mirror of https://github.com/systemd/systemd synced 2024-07-21 02:05:05 +00:00
Code Issues Actions 9 Packages Projects Releases Wiki Activity

tmpfiles: Try to take a BSD lock on files as well

Browse source 
Similar to what we do for directories, just before we remove a file,
let's try to take a BSD lock on it. If that fails, skip removing the
file.
This commit is contained in:
Daan De Meyer 2023-03-28 11:05:46 +02:00
parent 5ccf06074a
commit 65e179a1e7
4 changed files with 27 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

19
docs/TEMPORARY_DIRECTORIES.md
View file

@ -111,14 +111,13 @@ strategies to avoid these issues:
towards unexpected program termination as there are never files on disk that
need to be explicitly deleted.
3. 🥇 Operate below a sub-directory of `/tmp/` and `/var/tmp/` you created, and
take a BSD file lock ([`flock(dir_fd,
LOCK_SH)`](https://man7.org/linux/man-pages/man2/flock.2.html)) on that
sub-directory. This is particularly interesting when operating on more than
a single file, or on file nodes that are not plain regular files, for
example when extracting a tarball to a temporary directory. The ageing
algorithm will skip all directories (and everything below them) that are
locked through a BSD file lock. As BSD file locks are automatically released
3. 🥇 Take an exclusive or shared BSD file lock ([`flock()`](
https://man7.org/linux/man-pages/man2/flock.2.html)) on files and directories
you don't want to be removed. This is particularly interesting when operating
on more than a single file, or on file nodes that are not plain regular files,
for example when extracting a tarball to a temporary directory. The ageing
algorithm will skip all directories (and everything below them) and files that
are locked through a BSD file lock. As BSD file locks are automatically released
when the file descriptor they are taken on is closed, and all file
descriptors opened by a process are implicitly closed when it exits, this is
a robust mechanism that ensures all temporary files are subject to ageing
@ -127,9 +126,7 @@ strategies to avoid these issues:
modification/access times, as extracted files are otherwise immediately
candidates for deletion by the ageing algorithm. The
[`flock`](https://man7.org/linux/man-pages/man1/flock.1.html) tool of the
`util-linux` packages makes this concept available to shell scripts. Note
that `systemd-tmpfiles` only checks for BSD file locks on directories, locks
on other types of file nodes (including regular files) are not considered.
`util-linux` packages makes this concept available to shell scripts.
4. Keep the access time of all temporary files created current. In regular
intervals, use `utimensat()` or a related call to update the access time

4
man/systemd-tmpfiles.xml
View file

@ -131,7 +131,9 @@
directories marked with <varname>D</varname> or
<varname>R</varname>, and files or directories themselves
marked with <varname>r</varname> or <varname>R</varname> are
removed.</para></listitem>
removed unless an exclusive or shared BSD lock is taken on them (see <citerefentry
project='man-pages'><refentrytitle>flock</refentrytitle><manvolnum>2</manvolnum></citerefentry>).
</para></listitem>
</varlistentry>
<varlistentry>

12
man/tmpfiles.d.xml
View file

@ -647,13 +647,13 @@ w- /proc/sys/vm/swappiness - - - - 10</programlisting></para>
# an hour ago in "/tmp/foo/bar", are subject to time-based cleanup.
d /tmp/foo/bar - - - - bmA:1h -</programlisting></para>
<para>Note that while the aging algorithm is run a 'shared' BSD file lock (see <citerefentry
<para>Note that while the aging algorithm is run an exclusive BSD file lock (see <citerefentry
project='man-pages'><refentrytitle>flock</refentrytitle><manvolnum>2</manvolnum></citerefentry>) is
taken on each directory the algorithm descends into (and each directory below that, and so on). If the
aging algorithm finds a lock is already taken on some directory, it (and everything below it) is
skipped. Applications may use this to temporarily exclude certain directory subtrees from the aging
algorithm: the applications can take a BSD file lock themselves, and as long as they keep it aging of
the directory and everything below it is disabled.</para>
taken on each directory/file the algorithm decides to remove. If the aging algorithm finds a lock (
shared or exclusive) is already taken on some directory/file, it (and everything below it) is skipped.
Applications may use this to temporarily exclude certain directory subtrees from the aging algorithm:
the applications can take a BSD file lock themselves, and as long as they keep it aging of the
directory/file and everything below it is disabled.</para>
</refsect2>
<refsect2>

10
src/tmpfiles/tmpfiles.c
View file

@ -754,6 +754,8 @@ static int dir_cleanup(
r = log_warning_errno(errno, "Failed to remove directory \"%s\", ignoring: %m", sub_path);
} else {
_cleanup_close_ int fd = -EBADF;
/* Skip files for which the sticky bit is set. These are semantics we define, and are
* unknown elsewhere. See XDG_RUNTIME_DIR specification for details. */
if (sx.stx_mode & S_ISVTX) {
@ -794,6 +796,14 @@ static int dir_cleanup(
cutoff_nsec, sub_path, age_by_file, false))
continue;
fd = xopenat(dirfd(d), de->d_name, O_RDONLY|O_CLOEXEC|O_NOFOLLOW|O_NOATIME, 0);
if (fd < 0 && fd != -ENOENT)
log_warning_errno(fd, "Opening file \"%s\" failed, ignoring: %m", sub_path);
if (fd >= 0 && flock(fd, LOCK_EX|LOCK_NB) < 0 && errno == EAGAIN) {
log_debug_errno(errno, "Couldn't acquire shared BSD lock on file \"%s\", skipping: %m", p);
continue;
}
log_debug("Removing \"%s\".", sub_path);
if (unlinkat(dirfd(d), de->d_name, 0) < 0)
if (errno != ENOENT)