Persistent memory devices can be exposed as block devices as /dev/pmemN
and /dev/pmemNs. pmemN is the raw device and is byte-addressable from
within the kernel and when mmapped by applications from a DAX-mounted
file system. pmemNs has the block translation table (BTT) layered on top,
offering atomic sector/block access. Both pmemN and pmemNs are expected
to contain file systems.
blkid(8) and lsblk(8) seem to correctly report on pmemN and pmemNs.
systemd v219 will populate /dev/disk/by-uuid/ when, for example, mkfs is
used on pmem, but systemd v228 does not.
Add pmem to the whitelist.
The unit load queue can be processed in the middle of setting the
unit's properties, so its load_state would no longer be UNIT_STUB
for the check in bus_unit_set_properties(), which would cause it to
incorrectly return an error.
Commit da4d897e ("core: add cgroup memory controller support on the unified
hierarchy (#3315)") changed the code in src/core/cgroup.c to always write
the real numeric value from the cgroup parameters to the
"memory.limit_in_bytes" attribute file.
For parameters set to CGROUP_LIMIT_MAX, this results in the string
"18446744073709551615" being written into that file, which is UINT64_MAX.
Before that commit, CGROUP_LIMIT_MAX was special-cased to the string "-1".
This causes a regression on CentOS 7, which is based on kernel 3.10, as the
value is interpreted as *signed* 64 bit, and clamped to 0:
[root@n54 ~]# echo 18446744073709551615 >/sys/fs/cgroup/memory/user.slice/memory.limit_in_bytes
[root@n54 ~]# cat /sys/fs/cgroup/memory/user.slice/memory.limit_in_bytes
0
[root@n54 ~]# echo -1 >/sys/fs/cgroup/memory/user.slice/memory.limit_in_bytes
[root@n54 ~]# cat /sys/fs/cgroup/memory/user.slice/memory.limit_in_bytes
9223372036854775807
Hence, all units that are subject to the limits enforced by the memory
controller will crash immediately, even though they have no actual limit
set. This happens to for the user.slice, for instance:
[ 453.577153] Hardware name: SeaMicro SM15000-64-CC-AA-1Ox1/AMD Server CRB, BIOS Estoc.3.72.19.0018 08/19/2014
[ 453.587024] ffff880810c56780 00000000aae9501f ffff880813d7fcd0 ffffffff816360fc
[ 453.594544] ffff880813d7fd60 ffffffff8163109c ffff88080ffc5000 ffff880813d7fd28
[ 453.602120] ffffffff00000202 fffeefff00000000 0000000000000001 ffff880810c56c03
[ 453.609680] Call Trace:
[ 453.612156] [<ffffffff816360fc>] dump_stack+0x19/0x1b
[ 453.617324] [<ffffffff8163109c>] dump_header+0x8e/0x214
[ 453.622671] [<ffffffff8116d20e>] oom_kill_process+0x24e/0x3b0
[ 453.628559] [<ffffffff81088dae>] ? has_capability_noaudit+0x1e/0x30
[ 453.634969] [<ffffffff811d4155>] mem_cgroup_oom_synchronize+0x575/0x5a0
[ 453.641721] [<ffffffff811d3520>] ? mem_cgroup_charge_common+0xc0/0xc0
[ 453.648299] [<ffffffff8116da84>] pagefault_out_of_memory+0x14/0x90
[ 453.654621] [<ffffffff8162f4cc>] mm_fault_error+0x68/0x12b
[ 453.660233] [<ffffffff81642012>] __do_page_fault+0x3e2/0x450
[ 453.666017] [<ffffffff816420a3>] do_page_fault+0x23/0x80
[ 453.671467] [<ffffffff8163e308>] page_fault+0x28/0x30
[ 453.676656] Task in /user.slice/user-0.slice/user@0.service killed as a result of limit of /user.slice/user-0.slice/user@0.service
[ 453.688477] memory: usage 0kB, limit 0kB, failcnt 7
[ 453.693391] memory+swap: usage 0kB, limit 9007199254740991kB, failcnt 0
[ 453.700039] kmem: usage 0kB, limit 9007199254740991kB, failcnt 0
[ 453.706076] Memory cgroup stats for /user.slice/user-0.slice/user@0.service: cache:0KB rss:0KB rss_huge:0KB mapped_file:0KB swap:0KB inactive_anon:0KB active_anon:0KB inactive_file:0KB active_file:0KB unevictable:0KB
[ 453.725702] [ pid ] uid tgid total_vm rss nr_ptes swapents oom_score_adj name
[ 453.733614] [ 2837] 0 2837 11950 899 23 0 0 (systemd)
[ 453.741919] Memory cgroup out of memory: Kill process 2837 ((systemd)) score 1 or sacrifice child
[ 453.750831] Killed process 2837 ((systemd)) total-vm:47800kB, anon-rss:3188kB, file-rss:408kB
Fix this issue by special-casing the UINT64_MAX case again.
By cleaning up before setting up PAM we maintain control of overriding
behavior in setting variables. Otherwise, pam_putenv is in control.
This also makes sure we use a cleaned up environment in replacing
variables in argv.
Commit d054f0a4 ("tree-wide: use xsprintf() where applicable") used a
semantic patch approach to change a number of locations from
snprintf(buf, sizeof(buf), FMT, ...)
to
xsprintf(buf, FMT, ...)
The problem is that xsprintf() wraps the snprintf() in an
assert_message_se(), so if snprintf() reports an overflow of the
destination buffer, the binary will now terminate.
This hit a user running a version of systemd that was built from a
deeply nested system path.
Fix this by
a) Switching back to snprintf() for this particular case. We should really
rather truncate the location string than crash in such situations.
b) Increasing the size of that static string buffer, to make the event more
unlikely.
==1447== 4 bytes in 1 blocks are definitely lost in loss record 1 of 1
==1447== at 0x4C2BBAD: malloc (vg_replace_malloc.c:299)
==1447== by 0x5350F19: strdup (in /usr/lib64/libc-2.23.so)
==1447== by 0x4E9D435: strv_new_ap (strv.c:166)
==1447== by 0x4E9D5FA: strv_new (strv.c:199)
==1447== by 0x10E665: test_strv_fnmatch (test-strv.c:693)
==1447== by 0x10EAD5: main (test-strv.c:763)
==1447==
There are some places in the systemd which are use the same pattern:
fd_cloexec(STDIN_FILENO, false);
fd_cloexec(STDOUT_FILENO, false);
fd_cloexec(STDERR_FILENO, false);
to unset CLOEXEC for standard file descriptors. This patch introduces
the stdio_unset_cloexec() function to hide this and make code cleaner.
Add a new key ID_INPUT_TOUCHPAD_INTEGRATION=internal|external so we have a
single source for figuring out which touchpads are built-in.
Fairly simple approach: bluetooth is external, usb is external unless it's an
Apple touchpad. Everything else is internal.
Something has to so we can have udev rules rely on this. Right now the ID_BUS
setting is inconsistent: usb is set, ata and pci are set, bluetooth is not
set, rmi is too new to be featured.
70-mouse even relied on bluetooth even though it was never set
* networkd: condition_test() can return a negative error, handle that
If a condition check fails with an error we should not consider the check
successful. Fix that.
We should probably also improve logging in this case, but for now, let's just
unbreak this breakage.
Fixes: #3236
* condition: handle unrecognized architectures nicer
When we encounter a check for an architecture we don't know we should not
let the condition check fail with an error code, but instead simply return
false. After all the architecture might just be newer than the ones we know, in
which case it's certainly not our local one.
Fixes: #3236
This extends the existing event loop iteration counter to 64bit, and exposes it
via a new function sd_event_get_iteration(). This is helpful for cases like
issue #3612. After all, since we maintain the counter anyway, we might as well
expose it.
(This also fixes an unrelated issue in the man page for sd_event_wait() where
micro and milliseconds got mixed up)
By default, each iteration of manager_dispatch_sigchld() results in a unit level
sigchld event being invoked. For scope units, this results in a scope_sigchld_event()
which can seemingly stall for workloads that have a large number of PIDs within the
scope. The stall exhibits itself as a SIG_0 being initiated for each u->pids entry
as a result of pid_is_unwaited().
v2:
This patch resolves this condition by only paying to cost of a sigchld in the underlying
scope unit once per sigchld iteration. A new "sigchldgen" member resides within the
Unit struct. The Manager is incremented via the sd event loop, accessed via
sd_event_get_iteration, and the Unit member is set to the same value as the manager each
time that a sigchld event is invoked. If the Manager iteration value and Unit member
match, the sigchld event is not invoked for that iteration.
This extends the existing event loop iteration counter to 64bit, and exposes it
via a new function sd_event_get_iteration(). This is helpful for cases like
issue #3612. After all, since we maintain the counter anyway, we might as well
expose it.
(This also fixes an unrelated issue in the man page for sd_event_wait() where
micro and milliseconds got mixed up)
* test: check resolved generated resolv.conf in networkd-test
Directly verify the contents of /run/systemd/resolve/resolv.conf instead of
/etc/resolv.conf. The latter might be a plain file or a symlink to something
else (like Debian's resolvconf output), and in these cases we cannot make
strong assumptions about the contents.
Drop the "/etc/resolv.conf is a symlink" conditions and the "resolv.conf can
have at most three nameservers" alternatives, as we know that resolved always
adds all nameservers.
Explicitly start resolved at the start of a test to ensure that it is running.
* test: get along with existing system search domains in resolv.conf
The previous change has uncovered a bug in the tests: Existing search domains
can exist in resolv.conf which test_search_domains{,_too_long} didn't take into account.
As existing domains take some of the "max 6 domains" and "max 255 chars" limit,
don't expect that the last items from our test data actually appears in the
output, just the first few.
--boot=0 magically meant "this boot", but when used with --file/--directory it
should simply refer to the last boot found in the specified journal. This way,
--boot and --list-boots are consistent.
Fixes#3603.
Ivan Shapovalov