process_notifier() is called every time a process exits. When multiple
processes exit roughly at the same time, the uid_lock taken from inside
of process_notifier() will create contention which slows down process
exit. Defer stats accounting in such case to avoid lock contention.
Bug: 261537194
Change-Id: Ia1e9a451eab39eb0dda7eb175bfd71c67f3e0a58
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
A porting error in commit 0ae8088521d9 ("ANDROID: uid_cputime: Adds
accounting for the cputimes per uid.") caused
/proc/uid_cputime/show_uid_stat to report times in milliseconds
instead of microseconds. Restore the expected behavior.
Bug: 133273154
Bug: 146339763
Signed-off-by: Ed Savage-Jones <edward.savage-jones@sony.com>
Signed-off-by: Connor O'Brien <connoro@google.com>
Change-Id: I203919d1d44b300dab91810b16626a6950b063f8
(cherry picked from commit 305d5c197508b80a8dec0ab506de309e5d85fd26)
It's now being abstracted away, so fix up the ANDROID specific code that
touches the lock to use the correct functions instead.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I9cedb62118cd1bba7af27deb11499d312d24d7fc
Remove one more VLA that was not caught.
Bug: 154930031
Change-Id: I2c6c1f6d5e6ef23769c117a47689e1539b85c882
Signed-off-by: Alistair Delva <adelva@google.com>
This removes the portion of commit c408992c71 ("ANDROID: cpufreq:
times: track per-uid time in state") adding per-UID tracking, while
preserving an added null check in the per-task tracking code.
Bug: 127641090
Signed-off-by: Connor O'Brien <connoro@google.com>
Change-Id: Iae3e0abef32cbbb9fdc674079973040be0060ac7
proc files have their own filesystem ops type now, to make things
smaller and simpler. Fix up this code to handle this, otherwise it
breaks the build.
Signed-off-by: Greg Kroah-Hartman <gregkh@google.com>
Change-Id: I683c03a4744cb1356ba92bd5ed98adb395a3d1d5
When a thread is being killed process_notifier() is called to record
the final accounting of the thread. But after that uid_cputime_show()
and add_uid_io_stats() can be called before the dying thread is
removed from the parent's thread_group resulting in double accounting.
This can cause the user and system time for a given UID to move
backwards in /proc/uid_cputime/show_uid_stat. That gives negative delta
times in KernelCpuUidUserSysTimeReader.readDeltaImpl() and it logs
an error:
"Negative user/sys time delta for UID=..."
One consequence of which was incorrectly calculated power consumptions
in BatteryStats.
With this change we avoid the double accounting by ignoring the thread
if it has the PF_EXITING flag set.
Bug: 144366911
Change-Id: I6b929e8f558cd81ce1c00481c8b550d24877aa2c
Signed-off-by: Rickard Möller <rickard.moller@sony.com>
This adds a counter to the taskstats extended accounting fields, which
tracks the number of times fsync is called, and then plumbs it through
to the uid_sys_stats driver.
Bug: 120442023
Change-Id: I6c138de5b2332eea70f57e098134d1d141247b3f
Signed-off-by: Jin Qian <jinqian@google.com>
[AmitP: Refactored changes to align with changes from upstream commit
9a07000400 ("sched/headers: Move CONFIG_TASK_XACCT bits from <linux/sched.h> to <linux/sched/xacct.h>")]
Signed-off-by: Amit Pundir <amit.pundir@linaro.org>
[tkjos: Needed for storaged fsync accounting ("storaged --uid" and
"storaged --task").]
[astrachan: This is modifying a userspace interface and should probably
be reworked]
Signed-off-by: Alistair Strachan <astrachan@google.com>
IO usages are accounted in foreground and background buckets.
For each uid, io usage is calculated in two steps.
delta = current total of all uid tasks - previus total
current bucket += delta
Bucket is determined by current uid stat. Userspace writes to
/proc/uid_procstat/set <uid> <stat> when uid stat is updated.
/proc/uid_io/stats shows IO usage in this format.
<uid> <foreground IO> <background IO>
Bug: 34198239
Bug: 120442023
Change-Id: Ib8bebda53e7a56f45ea3eb0ec9a3153d44188102
Signed-off-by: Jin Qian <jinqian@google.com>
[connoro: Used by storaged.]
[astrachan: Note: this version does not track fsync syscalls; this is
implemented in a follow up due to the fact it depends on
modifications to the task_xacct feature.]
[astrachan: Folded in the following changes:
e003a91d8f42 ("ANDROID: uid_sys_stats: allow writing same state")
89b984bf2efd ("ANDROID: uid_sys_stats: fix negative write bytes.")
bb5ee21cc4dd ("ANDROID: uid_sys_stats: change to use rt_mutex")
9297d5a160c7 ("ANDROID: uid_sys_stats: reduce update_io_stats overhead")
89402d07fe91 ("ANDROID: uid_sys_stats: defer io stats calulation for dead tasks")
6dc5d8173a8c ("ANDROID: uid_sys_stats: check previous uid_entry before call find_or_register_uid")
0ca2ece8f7ec ("ANDROID: uid_sys_stats: log task io with a debug flag")
c9c096ef0e67 ("ANDROID: uid_sys_stats: Copy task_struct comm field to bigger buffer")]
Signed-off-by: Alistair Strachan <astrachan@google.com>
Adds proc files /proc/uid_cputime/show_uid_stat and
/proc/uid_cputime/remove_uid_range.
show_uid_stat lists the total utime and stime for the active as well as
terminated processes for each of the uids.
Writing a range of uids to remove_uid_range will delete the accounting
for all the uids within that range.
Bug: 120442023
Change-Id: I21d9210379da730b33ddc1a0ea663c8c9d2ac15b
Signed-off-by: Jin Qian <jinqian@google.com>
[AmitP: Refactored the original patch because upstream commit
605dc2b31a ("tsacct: Convert obsolete cputime type to nsecs")
made cputime_t type obsolete, so use u64 nanoseconds directly instead.
Also folded following android-4.9 changes into this patch
48a9906c0fd8 ("ANDROID: proc: uid_cputime: create uids from kuids")
453ac31cab34 ("ANDROID: proc: uid_cputime: fix show_uid_stat permission")]
Signed-off-by: Amit Pundir <amit.pundir@linaro.org>
[connoro: Used by batterystats.]
[astrachan: Folded in the file/kconfig rename and the following changes:
da3687fbc393 ("ANDROID: uid_cputime: fix cputime overflow")
71e3eb512d00 ("ANDROID: uid_cputime: Iterates over all the threads instead of processes.")
c65f0080c300 ("ANDROID: uid_cputime: Check for the range while removing range of UIDs.")
c8212458e3a5 ("ANDROID: uid_sys_stats: fix access of task_uid(task)")
bb5ee21cc4dd ("ANDROID: uid_sys_stats: change to use rt_mutex")
6dc5d8173a8c ("ANDROID: uid_sys_stats: check previous uid_entry before call find_or_register_uid")
82b9872b3894 ("ANDROID: uid_sys_stats: Replace tasklist lock with RCU in uid_cputime_show")]
Signed-off-by: Alistair Strachan <astrachan@google.com>
