Use the autogenerated libstatslog_lmkd to send events to statsd
The logging schema for statsd is changing as part of statsd becoming
a Mainline module in R. The autogenerated code will handle the schema
change.
Bug: 145887874
Test: m -j
Test: atest android.cts.statsd.atom.UidAtomTests#testLmkKillOccurred
Change-Id: Ibae4cd822807369a799d5c1f6a9c51272e38a074
lmkd unsolicited notifications can cause lmkd to block if clients are not
consuming them. Fix that by sending notifications to only subscribed
clients. Introduce LMK_SUBSCRIBE command to allow lmkd clients to subscribe
to event notifications. The only asynchronous event currently supported is
LMK_ASYNC_EVENT_KILL.
Bug: 146597855
Test: fill up send buffer using lmkd_unit_test
Test: confirm lmkd does not block after the fix
Change-Id: I014159aa55b59081f4b9ed53ecd160a49c0682bb
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
This change is to align how we get proc_name
between lmkd and lmkd_unit_test or it will always
fail when we run command such as ./lmkd_unit_test
under the device shell.
Bug: 146415484
Test: Verified using lmkd_unit_test
Change-Id: I3a6ebf0747a1054c5758b91fbacf3179b66b56c3
For now the only unsolicited message from lmkd is the process
kills on memory pressure.
Bug: 136036078
Test: atest ApplicationExitInfoTest
Change-Id: I503fd6a45ebab5276460b0ab978ebb2b8431dc0d
Signed-off-by: Jing Ji <jji@google.com>
pidhash is defined as an array of pointers:
static struct proc** pidhash = NULL;
...So we should be allocating `LINE_MAX * sizeof(struct proc *)` elems
here. Given the current constants here, this saves ~130KB, so not a big
deal, but still convenient.
Caught by clang's static analyzer:
system/memory/lmkd/statslog.c:354:19: warning: Result of 'calloc' is
converted to a pointer of type 'struct proc *', which is incompatible
with sizeof operand type 'struct proc'
[clang-analyzer-unix.MallocSizeof]
Bug: None
Test: TreeHugger
Change-Id: Iee9ca00a3a2a0ecababe9810d2ffcfc42169dd25
Add an optional process type field into lmkd registration protocol so that
applications can be distinguished from services.
Bug: 129011369
Test: boot and verify native service registration
Change-Id: Ie610b5d07cbe247a55ab31bc079ee5c5923bea11
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Introduce lmkd_unregister_proc helper function. Fix a bug where
lmkd_pack_set_procremove used a wrong structure as a parameter.
Bug: 129011369
Test: verify process record removal when it is manually killed
Change-Id: I7ab5a499f6b1c6eecfdba4d0a5ec916053e2726a
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
In order to register native services init needs ability to communicate with
lmkd. Make liblmkd_utils library available in recovery mode so that init
can link to it and add a data socket in lmkd to support additional
connection from init. Ensure SOCK_CLOEXEC type for lmkd socket to prevent
init children from inheriting it.
Bug: 129011369
Test: boot and verify native service registration
Change-Id: Iaa4f59282fb10f838f6811571e97d55754b1bd41
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Associate each registered process with the PID of the lmkd client that
registered it to prevent one client from updating records of another
client.
Bug: 129011369
Test: boot and verify native service registration
Change-Id: Id8ca7bb6314df225d04da6469b523d2cdc237eaa
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
We want poll_handler to be handler_info, so it's more efficient
to just unconditionally assign it.
Test: TreeHugger
Change-Id: I55b5164da1817ef77b5d455eb618f9a2471afc5c
lmkd uses PIDs to track processes, however occasionally a PID of a process
might be reused without lmkd detecting that. This can happen if originally
registered process crashes, PID numbers wrap around and the same PID gets
reused for a different process. In this situation lmkd might kill a wrong
process. To prevent this issue from occurring lmkd will track processes
using their pidfd. During process registration lmkd calls sys_pidfd_open
and stores returned pidfd with the process record. Returned pidfd will not
be reused until lmkd closes it which happens only after the process is
unregistered. This way lmkd ensures that process identification is unique
and can't be reused.
Bug: 135608568
Test: lmkd_unit_test with and without pidfd kernel support
Change-Id: Ida10ea13905c250e47f792cdd6bd2e65aeaa3709
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
With pidfd polling support lmkd can detect process death without periodic
polling. Implement mechanism to detect kernel pidfd support using
pidfd_open syscall existence as an indicator. Implement the logic to use
pidfd to wait for process death.
Bug: 135608568
Test: lmkd_unit_test with and without pidfd kernel support
Change-Id: Ic6db7e50893534467f5130a7f998b66fb4451272
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Caught by clang's static analyzer:
system/core/lmkd/lmkd.c:930:9: warning: 1st function call argument is an
uninitialized value [clang-analyzer-core.CallAndMessage]
Bug: None
Test: TreeHugger
Change-Id: I9dc8e97d6aa22ea977fa06553d957a31a9df8819
meminfo_log is used to log the state of the memory at the time of a kill.
Instead of reporting kill information and meminfo separately let's combine
them into one killinfo_log report. While normal logs can be trimmed by
chatty, meminfo_log uses a separate log context which gives it a better
chance of survival. As a result we will have all the information relevant
to a kill in one report which has higher chance of surviving chatty.
Bug: 74119935
Test: lmkd_unit_test
Change-Id: I83a9c12d538e1fb107721b04fdafc3c6c0d83b60
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Move statsd related code out of lmkd.c to minimize ifdefs sprinkled around
the code and make it more maintainable.
Bug: 74119935
Test: lmkd_unit_test
Change-Id: Ib22f90fd380b9a31e09ab18ef16787bc07415ddf
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
When lmkd fails to kill it should log the error, remove the process record
and exit immediately.
Bug: 74119935
Test: lmkd_unit_test
Change-Id: I26b0fd873eeed325f7dd56097ec31abc0d63f3a1
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Xin Li
Muhammad Qureshi
Automerger Merge Worker
Suren Baghdasaryan
Martin Liu
Tom Cherry
Jing Ji
George Burgess IV
Greg Kaiser
Treehugger Robot