Fix memory-safety and correctness bugs surfaced by Coverity audit (part 3) - #22268

…rt 3) (#22268)

* xenstat: skip domains without libxl domain info

Coverity CID 413854 (UNINIT): skip the current domain when `libxl_domain_info()` fails before hashing or storing its UUID. This avoids using an uninitialized `uuid` buffer on the error path in `xenstat_collect()`.

* datetime: fix rfc3339 fractional scaling

Coverity CID 457729 (INTEGER_OVERFLOW): print_fraction() underflowed its loop bound when callers requested 7-9 fractional digits.
Scale microseconds down for shorter output and pad zeros for 7-9 digits so the formatter preserves the existing 1..9-digit contract without hanging.

* libnetdata: grow cpuset cpu parser buffer

Coverity CID 425864 (TAINTED_SCALAR): os_read_cpuset_cpus() sized its static buffer from the first caller's system_cpus argument, and startup can first call it with 0. Derive a non-zero CPU baseline when needed and grow the buffer before reading cpuset.cpus so long CPU lists are not truncated into wrong counts or out-of-bounds parsing.

* proc: fix mdstat obsolete chart lookup key mismatch

Chart creation in proc_mdstat.c builds the short chart id "<raid>_<suffix>"
and passes it to rrdset_create_localhost("mdstat", id, ...); the RRD layer
later prefixes "mdstat." when constructing the full chart id.

make_chart_obsolete() was instead formatting the full "mdstat.<raid>_<suffix>"
directly into a 50-byte buffer and calling rrdset_find_active_byname_localhost,
which means the two paths used different truncation boundaries. For long but
valid md names (mdadm(8) allows up to 32 characters), the create-path chart
was longer than what obsolete-path could build, so the lookup missed and the
array's "availability" chart was never marked obsolete.

Build the same short chart id in the obsolete path and resolve it through
the type/id lookup helper so long valid md names match the created chart.

Related: Coverity CID 414643 flagged an OVERRUN on this line; the trace is
a tool-model false positive (no OOB in the current code), but investigating
it surfaced the real correctness bug fixed here.

* proc.plugin: avoid uaf in power supply property loop

Coverity CID 348628 (USE_AFTER_FREE): do_sys_class_power_supply()
could free the current power supply while iterating its property list,
then evaluate the outer loop increment on the freed property node.
Store the next property before the inner loop and stop iterating once
the error path frees the power supply.

* rrdhost: fix unlocked receiver status snapshot

Coverity CID 410067 (MISSING_LOCK): rrdhost_status_ingest() read receiver
status fields without receiver_lock while the receiver thread updated the same
state under that lock. Snapshot the receiver status block under the lock before
deriving ingest status so status reporting no longer races with connect and
disconnect updates.

* aclk: fix pending request cancellation race

Coverity CID 410087 (MISSING_LOCK): aclk_web_client_interrupt_cb() read pending_req_list.canceled from a libuv worker while cancel paths updated the same flag from the ACLK side.
Switch the cancel flag accesses to atomic load/store so query cancellation stays lock-free on the hot callback path without racing across threads.

* statsd: lock histogram sample buffer access

Coverity CID 410124 (MISSING_LOCK): histogram and timer samples updated
the shared buffer partly outside `m->histogram.ext->mutex` while the flush
path sorted the same buffer under that lock. Keep reset, growth, append, and
flush snapshot on the same mutex so samples are not dropped or read from a
partially synchronized buffer.

* apps.plugin: unlock mutex before print exit

Coverity CID 381151 (LOCK): the stored trace points to stale line numbers, but
the current print-tree exit path still called exit() while holding
apps_and_stdout_mutex. Unlock the mutex after printing and before exit() so the
module destructor does not destroy a locked uv_mutex_t.

---------

Co-authored-by: Costa Tsaousis <costa@netdata.cloud>
(cherry picked from commit eee9c55)