| Age | Commit message (Collapse) | Author |
|---|
|
We were compressing unitialized memory, which should not result in
any problems, but is inelegant.
|
|
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.
|
|
Logging for compression and decompression is assymetrical on purpose:
if compiled without some type of compression, those compression code
paths should never be invoked. OTOH, it is possible to encounter
unsupported format on decompression, so leave those log_debug statements
in, to make it easier to diagnose stuff.
|
|
|
|
Existing test would use highly-compressible repeatable
input. Two types of input are added:
- zeros
- random blocks interspersed with zeros
The idea is to get more information about behaviour in various cases.
On Intel Xeon the results are:
% ./test-compress-benchmark
XZ/zeros: compressed & decompressed 2535301373 bytes in 32.56s (74.26MiB/s), mean compresion 99.96%, skipped 3160 bytes
LZ4/zeros: compressed & decompressed 2535304362 bytes in 1.16s (2088.69MiB/s), mean compresion 99.60%, skipped 171 bytes
XZ/simple: compressed & decompressed 2535300963 bytes in 30.42s (79.48MiB/s), mean compresion 99.95%, skipped 3570 bytes
LZ4/simple: compressed & decompressed 2535303543 bytes in 1.22s (1978.86MiB/s), mean compresion 99.60%, skipped 990 bytes
XZ/random: compressed & decompressed 381756649 bytes in 60.02s (6.07MiB/s), mean compresion 39.64%, skipped 27813723 bytes
LZ4/random: compressed & decompressed 2507385036 bytes in 0.97s (2477.52MiB/s), mean compresion 54.77%, skipped 27919497 bytes
If someone has ideas for more realistic test cases, they can be easily
added to this framework.
|
|
This converts the stream compression to use the new lz4frame api,
compatible with lz4cat. Previous code used custom headers, so the
compressed file was not compatible with lz4 command line tools.
I considered this the last blocker to using lz4 by default.
Speed seems to be reasonable, although a bit (a few percent) slower
than the lz4 binary, even though compression is the same. I don't
consider this important. It could be caused by the overhead of library
calls, but is probably caused by slightly different buffer sizes or
such. The code in this patch uses mmap, since since this allows the
buffer to be reused while not making the code more complicated at all.
In my testing, this version is noticably faster (~20%) than a naive
single-buffered version. mmap can cause the program to be killed with
SIGBUS, if the underlying file is truncated or a disk error occurs. We
only use this from within coredump and coredumpctl, so I don't
consider this an issue.
Old decompression code is retained and is used if the new code fails
indicating a format error. There have been reports of various smaller
distributions using previous lz4 code, i.e. the old format, and it is
nice to provide backwards compatibility. We can remove the legacy code
in a few versions.
The way that blobs are compressed in the journal is not affected.
|
|
off_t is a really weird type as it is usually 64bit these days (at least
in sane programs), but could theoretically be 32bit. We don't support
off_t as 32bit builds though, but still constantly deal with safely
converting from off_t to other types and back for no point.
Hence, never use the type anymore. Always use uint64_t instead. This has
various benefits, including that we can expose these values directly as
D-Bus properties, and also that the values parse the same in all cases.
|
|
Third round of Coccinelle fixes
|
|
|
|
And set_free() too.
Another Coccinelle patch.
|
|
Another Coccinelle patch.
|
|
logind: make dry run command line arguments work again (v2)
|
|
Prefix the action parameter with "dry-" in case the --dry-run command
line switch was passed.
|
|
Allow passing a "dry-" prefix to the action parameter passed to
.ScheduleShutdown(). When strings with this prefix are passed,
the scheduled action will not take place. Instead, an info message
is logged.
|
|
notify: log error when sd_pid_notify() == 0
|
|
util: introduce safe_fclose() and port everything over to it
|
|
Adds a coccinelle script to port things over automatically.
|
|
locale: kill free_and_replace()
|
|
That function really makes little sense, as the open-coded variant
is much more readable. Also, if the 2nd argument is NULL, mfree()
is a much better candidate.
Convert the only users of this function in localed, and then remove it
entirely.
|
|
Coccinelle fixes 2
|
|
Another Coccinelle script.
|
|
Let's also clean up single-line while and for blocks.
|
|
The previous coccinelle semantic patch that improved usage of
log_error_errno()'s return value, only looked for log_error_errno()
invocations with a single parameter after the error parameter. Update
the patch to handle arbitrary numbers of additional arguments.
|
|
We should never put empty lines between `if` and `else if`, unless we use
braces.
|
|
sd_pid_notify_with_fds: fix computing msg_controllen
|
|
core: freeze execution if /etc/mtab exists
|
|
CMSG_SPACE(0) may return value other than 0. This caused sendmsg to fail
with EINVAL, when have_pid or n_fds was 0.
|
|
The mount monitor that was added to libmount v2.27 requires /etc/mtab to be
non-existant. As systemd now uses that functionality, we cannot monitor any
mounts anymore, and hence not support .mount units.
Systems that have /etc/mtab around as regular file are unsupported by
systemd since a long time.
This patch makes that condition fatal, so we do not boot up with
non-working mount monitor support.
|
|
|
|
Even though systemd has its own smack label since
'--with-smack-run-label' configuration is set, the smack label of each
CGROUP root directory should have the star (i.e. *) label. This is
mainly because current Linux Kernel set the label in this way.
(Refer to smack_d_instantiate() in security/smack/smack_lsm.c)
However, if systemd has its own smack label and arg_join_controllers is
explicitly set or initialized by initialize_join_controllers() function,
current systemd creates the symlink in CGROUP root directory with its
own smack label as below.
lrwxrwxrwx. 1 root root System 11 Dec 31 16:00 cpu -> cpu,cpuacct
dr-xr-xr-x. 4 root root * 0 Dec 31 16:01 cpu,cpuacct
lrwxrwxrwx. 1 root root System 11 Dec 31 16:00 cpuacct -> cpu,cpuacct
This patch fixes that bug by copying the smack label from the origin.
|
|
This adds a new mac_smack_copy() function in order to read the smack
label from the source and apply it to the destination.
|
|
|
|
cgroup fix, nspawn fix, plus change to download .nspawn files in importd
|
|
Coccinelle fixes
|
|
|
|
analyze: add alias handling for --{from,to}-pattern options of the dot command
|
|
Patch via coccinelle.
|
|
Turns this:
r = -errno;
log_error_errno(errno, "foo");
into this:
r = log_error_errno(errno, "foo");
and this:
r = log_error_errno(errno, "foo");
return r;
into this:
return log_error_errno(errno, "foo");
|
|
Turn this:
if ((r = foo()) < 0) { ...
into this:
r = foo();
if (r < 0) { ...
|
|
Replace this:
if (fd >= 0)
safe_close(fd);
by this:
safe_close(fd);
|
|
Replace this:
close(fd);
fd = -1;
write this:
fd = safe_close(fd);
|
|
This replaces this:
free(p);
p = NULL;
by this:
p = mfree(p);
Change generated using coccinelle. Semantic patch is added to the
sources.
|
|
udev: build by-path identifiers for ATA devices.
|
|
/dev/disk/by-path
total 0
lrwxrwxrwx. 1 root root 9 Sep 4 10:02 pci-0000:00:1f.2-ata-2 -> ../../sr0
lrwxrwxrwx. 1 root root 9 Sep 4 10:02 pci-0000:00:1f.2-ata-3 -> ../../sdd
lrwxrwxrwx. 1 root root 10 Sep 4 10:02 pci-0000:00:1f.2-ata-3-part1 -> ../../sdd1
lrwxrwxrwx. 1 root root 10 Sep 4 10:02 pci-0000:00:1f.2-ata-3-part2 -> ../../sdd2
lrwxrwxrwx. 1 root root 10 Sep 4 10:02 pci-0000:00:1f.2-ata-3-part3 -> ../../sdd3
lrwxrwxrwx. 1 root root 9 Sep 4 10:02 pci-0000:03:00.0-ata-4 -> ../../sda
lrwxrwxrwx. 1 root root 10 Sep 4 10:02 pci-0000:03:00.0-ata-4-part1 -> ../../sda1
lrwxrwxrwx. 1 root root 10 Sep 4 10:02 pci-0000:03:00.0-ata-4-part2 -> ../../sda2
lrwxrwxrwx. 1 root root 9 Sep 4 10:02 pci-0000:08:00.0-ata-1 -> ../../sdc
lrwxrwxrwx. 1 root root 10 Sep 4 10:02 pci-0000:08:00.0-ata-1-part1 -> ../../sdc1
lrwxrwxrwx. 1 root root 10 Sep 4 10:02 pci-0000:08:00.0-ata-1-part2 -> ../../sdc2
|
|
Since 59512f21 the parameters were passed in the wrong order, causing the
options to be interpreted as the description.
Also, while "false" was supposed to be passed for "rw", the ESP should actually
be mounted read-write. It just happened to be "true" since the description char*
was passed for "rw".
|
|
|
|
the controller in the kernel
Follow-up to 5bf8002a3a6723ce50331c024122078552fb600a.
|
|
basic: rework virtualization detection API
|
|
This seems to be an oversight from:
707b66c66381c899d7ef640e158ffdd5bcff4deb
We have to return ENODATA instead of ENOENT if a requested entry is
non-present. Also fix the call-site in udev to check for these errors.
|
|
treewide: fix typos
|
