If a process forked and the child process was killed by the
lowmemorykiller, the lowmemory killer would be disabled until
the parent process reaped the child or it died itself.
Change-Id: I709b1a4e1b1a1970e51d26a39fcbee57977bbc7f
Signed-off-by: Arve Hjønnevåg <arve@android.com>
The lowmemorykiller registers an atomic notifier for notfication of when
the task is freed. From this atomic notifier callback, it removes the
atomic notifier via task_free_unregister(). This is incorrect because
atomic_notifier_chain_unregister() calls syncronize_rcu(), which can
sleep, which shouldn't be done from an atomic notifier.
Fix this by registering the notifier during init, and only unregister it
if the lowmemorykiller is unloaded.
Change-Id: I1577b04e617bc2b2e39dcb490fcfc9ce660eb7ec
Signed-off-by: Rabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: Christian Bejram <christian.bejram@stericsson.com>
synchronize_rcu can be very expensive, averaging 100 ms in
some cases. In cgroup_attach_task, it is used to prevent
a task->cgroups pointer dereferenced in an RCU read side
critical section from being invalidated, by delaying the
call to put_css_set until after an RCU grace period.
To avoid the call to synchronize_rcu, make the put_css_set
call rcu-safe by moving the deletion of the css_set links
into free_css_set_work, scheduled by the rcu callback
free_css_set_rcu.
The decrement of the cgroup refcount is no longer
synchronous with the call to put_css_set, which can result
in the cgroup refcount staying positive after the last call
to cgroup_attach_task returns. To allow the cgroup to be
deleted with cgroup_rmdir synchronously after
cgroup_attach_task, have rmdir check the refcount of all
associated css_sets. If cgroup_rmdir is called on a cgroup
for which the css_sets all have refcount zero but the
cgroup refcount is nonzero, reuse the rmdir waitqueue to
block the rmdir until free_css_set_work is called.
Signed-off-by: Colin Cross <ccross@android.com>
Changes the meaning of CGRP_RELEASABLE to be set on any cgroup
that has ever had a task or cgroup in it, or had css_get called
on it. The bit is set in cgroup_attach_task, cgroup_create,
and __css_get. It is not necessary to set the bit in
cgroup_fork, as the task is either in the root cgroup, in
which can never be released, or the task it was forked from
already set the bit in croup_attach_task.
Signed-off-by: Colin Cross <ccross@android.com>
If the parent device is in the DPM_RESUMING state, resume might get called
before the parent's resume has completed, because dpm_wait won't get
called. This issue was introduced by:
3d46a3c PM: Prevent waiting forever on asynchronous resume after abort
Wait for completion of the parent resume if the parent state is >= DPM_OFF
or == DPM_RESUMING
Signed-off-by: Benoit Goby <benoit@android.com>
[dima: fixed to work on 2.6.38, use .in_suspend instead of .state]
Signed-off-by: Dima Zavin <dima@android.com>
Need to use irqsave/restore spin locking for cpumasks since these
are accessed in timers and in thread context.
Change-Id: I4a53eaf0ced7e73b445feddba90ec11482de9126
Signed-off-by: Todd Poynor <toddpoynor@google.com>
This governor is designed for latency-sensitive workloads, such as
interactive user interfaces. The interactive governor aims to be
significantly more responsive to ramp CPU quickly up when CPU-intensive
activity begins.
Existing governors sample CPU load at a particular rate, typically
every X ms. This can lead to under-powering UI threads for the period of
time during which the user begins interacting with a previously-idle system
until the next sample period happens.
The 'interactive' governor uses a different approach. Instead of sampling
the CPU at a specified rate, the governor will check whether to scale the
CPU frequency up soon after coming out of idle. When the CPU comes out of
idle, a timer is configured to fire within 1-2 ticks. If the CPU is very
busy from exiting idle to when the timer fires then we assume the CPU is
underpowered and ramp to MAX speed.
If the CPU was not sufficiently busy to immediately ramp to MAX speed, then
the governor evaluates the CPU load since the last speed adjustment,
choosing the highest value between that longer-term load or the short-term
load since idle exit to determine the CPU speed to ramp to.
A realtime thread is used for scaling up, giving the remaining tasks the
CPU performance benefit, unlike existing governors which are more likely to
schedule rampup work to occur after your performance starved tasks have
completed.
The tuneables for this governor are:
/sys/devices/system/cpu/cpufreq/interactive/min_sample_time:
The minimum amount of time to spend at the current frequency before
ramping down. This is to ensure that the governor has seen enough
historic CPU load data to determine the appropriate workload.
Default is 80000 uS.
/sys/devices/system/cpu/cpufreq/interactive/go_maxspeed_load
The CPU load at which to ramp to max speed. Default is 85.
Change-Id: Ib2b362607c62f7c56d35f44a9ef3280f98c17585
Signed-off-by: Mike Chan <mike@android.com>
Signed-off-by: Todd Poynor <toddpoynor@google.com>
Bug: 3152864
Adds a config option that controls whether or not the console is on at
boot, defaulting to no.
Change-Id: Id4a3cad1c9dd4fb3c2b8c2298ca5d385a8bd0f8d
Signed-off-by: Dima Zavin <dima@android.com>
To allow the set_clock host op to disable the SDCLK source when not
needed, always call the host op when the requested clock speed is
zero. Do this even if host->clock already equals zero, because
the SDHCI driver may set that value (without calling the host op)
to force an update at the next (non-zero-speed) call.
Change-Id: If99230d76138679b5767f77cb925f15408ae518e
Signed-off-by: Todd Poynor <toddpoynor@google.com>
Prints the number of local timer irqs on each cpu when
the irqs command is executed.
Change-Id: Ic70fa9f528d98996fabb2d0cc80e937c5b239c1a
Signed-off-by: Rebecca Schultz Zavin <rebecca@android.com>
This allows us to get a kernel stacktrace for a thread though /proc.
Also enable it by default.
Change-Id: If8c21cd02feaf9863f4841ace524fa30c7328d49
Signed-off-by: Arve Hjønnevåg <arve@android.com>
the csd sector count reported by eMMC 4.3+ cards includes the boot
partition size; subtract this from the size reported to the disk
since the boot partition is inaccessible
Change-Id: I601b83aa0159b7aa446409ea8c945b256dd0b5b1
Signed-off-by: Gary King <gking@nvidia.com>
Set the GENHD_FL_EXT_DEVT flag, which will allocate minor numbers
in major 259 for partitions past disk->minors.
Also remove the use of disk_devt to determine devidx from md->disk.
md->disk->first_minor is always initialized from devidx and can
always be used to recover it.
Signed-off-by: Colin Cross <ccross@android.com>
Only wait on a parent device during resume if the parent device is
suspended.
Consider three drivers, A, B, and C. The parent of A is C, and C
has async_suspend set. On boot, C->power.completion is initialized
to 0.
During the first suspend:
suspend_devices_and_enter(...)
dpm_resume(...)
device_suspend(A)
device_suspend(B) returns error, aborts suspend
dpm_resume_end(...)
dpm_resume(...)
device_resume(A)
dpm_wait(A->parent == C)
wait_for_completion(C->power.completion)
The wait_for_completion will never complete, because
complete_all(C->power.completion) will only be called from
device_suspend(C) or device_resume(C), neither of which is called
if suspend is aborted before C.
After a successful suspend->resume cycle, where B doesn't abort
suspend, C->power.completion is left in the completed state by the
call to device_resume(C), and the same call path will work if B
aborts suspend.
Signed-off-by: Colin Cross <ccross@android.com>
For android builds we disable the check for curlun->prevent_medium_removal.
Instead we let the framework manage unmounting policy, as we sometimes need
to unmount after the media has been removed.
This also helps support hosts that do not inform the device when the media
has been unmounted.
Signed-off-by: Mike Lockwood <lockwood@android.com>
Platform must register cpu power function that return power in
milliWatt seconds.
Change-Id: I1caa0335e316c352eee3b1ddf326fcd4942bcbe8
Signed-off-by: Mike Chan <mike@android.com>
Introduce new platform callback hooks for cpuacct for tracking CPU frequencies
Not all platforms / architectures have a set CPU_FREQ_TABLE defined
for CPU transition speeds. In order to track time spent in at various
CPU frequencies, we enable platform callbacks from cpuacct for this accounting.
Architectures that support overclock boosting, or don't have pre-defined
frequency tables can implement their own bucketing system that makes sense
given their cpufreq scaling abilities.
New file:
cpuacct.cpufreq reports the CPU time (in nanoseconds) spent at each CPU
frequency.
Change-Id: I10a80b3162e6fff3a8a2f74dd6bb37e88b12ba96
Signed-off-by: Mike Chan <mike@android.com>
Originally written by HTC. Contributions by Motorola and AKM.
misc: Import akm8975 from Motorola
Major style and code cleanups by Praveen Bharathi <pbharathi@motorola.com>
misc: akm8975: clean up code violations in akm8975.c
misc: akm8975: Clean up coding style, add suspend and resume
Change-Id: I4196913f15aec2dfbed47506d3dc085aada8e92d
Signed-off-by: Dima Zavin <dima@android.com>
When enabled, tracks the frequency of network transmissions
(inbound and outbound) and buckets them accordingly.
Buckets are determined by time between network activity.
Each bucket represents the number of network transmisions that were
N sec or longer apart. Where N is defined as 1 << bucket index.
This network pattern tracking is particularly useful for wireless
networks (ie: 3G) where batching network activity closely together
is more power efficient than far apart.
New file: /proc/net/stat/activity
output:
Min Bucket(sec) Count
1 7
2 0
4 1
8 0
16 0
32 2
64 1
128 0
Change-Id: I4c4cd8627b872a55f326b1715c51bc3bdd6e8d92
Signed-off-by: Mike Chan <mike@android.com>
Now that we're murder-synchronous, this code path will never be
called (and if it does, it doesn't tell us anything useful other
than we killed a task that was already being killed by somebody
else but hadn't gotten its' signal yet)
Signed-off-by: San Mehat <san@google.com>
As it turns out, the CONFIG_PROFILING interfaces leak a
task struct if the notifier chain returns NOTIFY_OK.. doh.
This patch reworks lowmemkiller to use the new generic task
free notifier chain.
Signed-off-by: San Mehat <san@google.com>
This patch adds a notifier which can be used by subsystems that may
be interested in when a task has completely died and is about to
have it's last resource freed.
The Android lowmemory killer uses this to determine when a task
it has killed has finally given up its goods.
Signed-off-by: San Mehat <san@google.com>
binder_deferred_release was not unmapping the page from the buffer
before freeing it, causing memory corruption. This only happened
when page(s) had not been freed by binder_update_page_range, which
properly unmaps the pages.
This only happens on architectures with VIPT aliasing.
To reproduce, create a program which opens, mmaps, munmaps, then closes
the binder very quickly. This should leave a page allocated when the
binder is released. When binder_deferrred_release is called on the
close, the page will remain mapped to the address in the linear
proc->buffer. Later, we may map the same physical page to a different
virtual address that has different coloring, and this may cause
aliasing to occur.
PAGE_POISONING will greatly increase your chances of noticing any
problems.
Change-Id: I6941bf212881b8bf846bdfda43d3609c7ae4892e
Signed-off-by: Christopher Lais <chris+android@zenthought.org>
This patch optimizes lowmemkiller to not do any work when it has an outstanding
kill-request. This greatly reduces the pressure on the task_list lock
(improving interactivity), as well as improving the vmscan performance
when under heavy memory pressure (by up to 20x in tests).
Note: For this enhancement to work, you need CONFIG_PROFILING
Signed-off-by: San Mehat <san@google.com>
