I am closing this master bug since all the referenced fixes have been made.
Idle rate is now sub 1% per hr drain on Krillin
These fixes are in the upcoming update
** Changed in: powerd (Ubuntu)
Status: Confirmed = Fix Released
** Changed in: powerd
Status: New = Fix Released
**
After checking the suspend logic during this weekend, I ended up with 2
remaining bugs that are probably helping battery drain:
Bug 1435109 - Push-client frequently blocking phone suspend
Bug 1434379 - GPS always active when a scope that uses location is in the
background
--
You received
Looking at the logs, it seems there are 3 patterns atm (on mako):
1 - Wifi driver failing to suspend frequently:
Feb 6 12:07:20 ubuntu-phablet kernel: [116497.506926] suspend: enter suspend
Feb 6 12:07:20 ubuntu-phablet kernel: [116497.506956] PM: suspend entry
2015-02-06 17:07:20.782768475
1 - For the wifi driver there is really not much to do, also an issue
with Android and it seems to be in the latest kernel tree from google as
well (need to properly check with krillin to see if a similar issue can
be found in there).
--
You received this bug notification because you are a
3 - Can't really reproduce when testing with my device, so created the
following MR to enable debug in powerd by default (safe as it's really
not printing much more because of that, but should give enough to find
out why our userspace is not even trying to suspend):
** Also affects: powerd
Importance: Undecided
Status: New
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery drain on RTM
Status in
ongoing so assigning to last milestone
** Changed in: canonical-devices-system-image
Milestone: ww07-2015 = ww13-2015
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
** Changed in: canonical-devices-system-image
Assignee: (unassigned) = Pat McGowan (pat-mcgowan)
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Any chance of getting that syslog to see what's going on?
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery drain on RTM
Status in the base
FYI, For my tests, I am reading the battery capacity directly, just to
keep it as light weight as possible:
#!/bin/sh
rm -f battery.log
while true
do
c=$(cat /sys/devices/platform/battery/power_supply/battery/capacity)
d=$(date +%d/%m/%Y %H:%M:%S)
echo $d $c battery.log
I've correlated each processes against the spike in activity for upowerd
and also see that mpdecision and systemd-udevd also change their
behaviors at that transition. See attached spreadsheet.
** Attachment added: upowerd-procs.ods
I noticed that at this point, the syslog shows and incoming call
occurred. After that, the phone does NOT deep suspend at all.
Eventually, the phone drains at 03:55:33 on the 7th of Feb. So I think
somebody needs to look at the way incoming calls seem to block deep
suspend.
Feb 6 17:12:05
So there is a rise of activity with systemd-udev at the same time that
upowerd gets busy too
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive
Pat, do you mind attaching the entire cpustat.log so I spot any specific
trends? Thanks
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery
Pat, I use something such like the following awk script to parse the
data:
{
if ((NF == 5) ($4 != PID)) {
total[$4] += $1
usr[$4] += $2
sys[$4] += $3
cmd[$4] = $5
}
if ((NF == 5) ($4 == PID))
We could possibly attach health-check to powerd to see what activity is
going.
E.g.
health-check -r -w -W -f -c -p upowerd -d 3600 health-check.log
..that will attach health-check for 1 hour and dump the results into
health-check.log.
Colin
--
You received this bug notification because you
The complete cpustat.log for that event
@colin do you have something to intelligently parse these files?
** Attachment added: cpustat.log
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4315423/+files/cpustat.log
--
You received this bug notification because you
and the syslog I somehow did't attach
** Attachment added: syslog
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4315438/+files/syslog
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in
I had a drainoff last night on mako and had cpustat running, which points at
upowerd
A number of samples like the following :
%CPU %USR %SYS PID Task
17.76 16.87 0.90 1493 /usr/lib/upower/upowerd
0.80 0.40 0.40 4319 cpustat
0.80 0.10 0.70 1180 /system/bin/mpdecision
Syslog showing lots of wmlan and mmc reports starting at 17:53 when I
left the house and out of wifi range.
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
** Tags added: battery
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery drain on RTM
Status in the base for Ubuntu mobile products:
In
(see also comment #5 above)
sudo add-apt-repository ppa:colin-king/white
sudo apt-get update
sudo apt-get install cpustat
cpustat blocks on a select() system call. While the phone is not suspended it
will log all CPU process activity every 2 seconds. This will consume ~0.25% of
the CPU, so in
Krillin with wifi and bt off
Lost 36.000 percent over 20.76 hours, 1.73 percent per hr
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery
Mako with wifi turned off
Lost 19.000 percent over 14.88 hours, 1.27 percent per hr
it was 10% drained after 13 hrs idle overnight
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
** Changed in: canonical-devices-system-image
Importance: Undecided = Critical
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery drain on
** Attachment added: Script to analyze the upower charge history dat file
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4309024/+files/dcstats
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to
I wrote a script that parses the contents of /var/lib/upower//history-
charge-generic_id.dat
The recent results on krillin:
Lost 43.000 percent over 18.33 hours, 2.34 percent per hr
Lost 5.000 percent over 1.69 hours, 2.94 percent per hr
Lost 58.000 percent over 15.97 hours, 3.62 percent per hr
In a dupe of this bug they report around 4% drain per hour (47% over 12
hrs)
I am seeing an average, including usage not just when idle, of 3.2% on krillin
and 3.1% on mako
Seeing as low as 2.3% when idle on krillin and as low as 1.3% when idle on
mako, all percent drop per hour.
Wifi, BT,
** Attachment added: krillin with less drain
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4307255/+files/krbatt2.jpg
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
** Attachment added: mako with excessive drain
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4307253/+files/makobatt2.jpg
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
** Attachment added: mako with excessive drain
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4307254/+files/makobatt2.jpg
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
** Changed in: canonical-devices-system-image
Milestone: ww05-2015 = ww07-2015
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery drain on
** Attachment added: Krillin unplugged overnight with excessive battery drain
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4306528/+files/krilinbatt.jpg
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is
** Attachment added: Mako unplugged overnight with reasonable battery drain
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4306527/+files/makobatt.jpg
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is
Results today on my Krillin, each sample is around an hour
(Mako shows 2% suspends aborted while associated.)
With no AP associated:
Suspend blocking wakelocks:
None
Resume wakeup causes:
CCIF_MD96 78.69%
EINT 26 21.31%
comments from Colin
I've worked through all the images now and can see that deep suspend
sometimes fails to work completely because of a series of COMM events
that cause excessive wakeups. Typically, in this scenario, one sees
300-700 CONN wakeup events over a 300 second interval. This either
With wifi associated to a fairly busy AP that produces regular beacon
intervals with CTS protection mode enabled and also phone data enabled
I'm seeing ~24+ hours on deep sleep idle.
With wifi enabled, I'm seeing ~7.5 wifi related wakeups per minute. I
then disabled wifi and these wakeups
I've written a bash script to periodically grab battery capacity levels
over some of this afternoon and projected the expected duration (making
an assumption the battery drain is linear, which it is not, i know
that). Anyhow, I predict 30+ hours of idle on a default clean install
on deep sleep.
I also compared 10 minutes of file process wakeup activity (old pre-
Christmas image vs lastest RTM image) and I'm seeing little different
between old vs new, so no obvious regressions there.
See attached files (spreadsheet and old vs new file activity logs).
** Attachment added:
It's weird that we got so much more activity on /lib/udev/rules.d,
/etc/modprobe.d and /etc/group, wonder what might be causing that.
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
I compared 10 minutes of file access activity (old pre-Christmas image
vs lastest RTM image) and I'm seeing a lot more udevd activity on
/lib/udev/rules.d, /etc/modprobe.d and /etc/group.
See attached files (spreadsheet and old vs new file activity logs).
** Attachment added: tar of old vs new
** Changed in: canonical-devices-system-image
Milestone: ww03-2015 = ww05-2015
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery drain on
** Changed in: canonical-devices-system-image
Status: Confirmed = In Progress
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
https://bugs.launchpad.net/bugs/1372413
Title:
Extensive battery drain
My phone battery was virtually flat again this morning.
I'll see if I can get someone to help with the tech stuff above - meanwhile
though here is a screen-grab of my battery discharge graph from settings which
clearly shows a continuous (and fairly rapid) discharge overnight despite
being,
Mako running 160 out of battery, lost 50% charge overnight
While it seems to suspend successfully the majority of the time it is not
suspended very long, 90% are less than a min
** Attachment added: sb-mako-160
** Attachment added: batt.jpg
https://bugs.launchpad.net/ubuntu/+source/powerd/+bug/1372413/+attachment/4299163/+files/batt.jpg
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
I compared a 10 minute sleep from an image from pre-Christmas to the
latest RTM image and I see better deep suspend stats with the latest
image (see attached).
If Thomas can attach all the syslog files from /var/log/syslog we can
parse this with suspend-blocker and get an idea of what's been
I compared a 10 minute sleep from an image from pre-Christmas to the
latest RTM image and I see better deep suspend stats with the latest
image (see attached).
If Thomas can attach all the syslog files from /var/log/syslog we can
parse this with suspend-blocker and get an idea of what's been
Reviving this bug based on recent anecdotal reports that battery life
has degraded
Folks who experience it can do several things to provide data:
1 - check the battery page in settings and see the battery level history, for
example did it discharge overnight
2 - provide the syslog which will
adding here to track
** Also affects: canonical-devices-system-image
Importance: Undecided
Status: New
** Changed in: canonical-devices-system-image
Status: New = Confirmed
** Changed in: canonical-devices-system-image
Milestone: None = ww03-2015
--
You received this bug
Also, it would be useful to capture activity of the system using the
following over a 10 minute period
Still using ppa:colin-king/white:
sudo apt-get install eventstat cpustat fnotifystat forkstat
sudo eventstat 10 60 eventstat.log
(10 seconds of sampling, 60 x a second)
And see if the CPU
Status changed to 'Confirmed' because the bug affects multiple users.
** Changed in: powerd (Ubuntu)
Status: New = Confirmed
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to powerd in Ubuntu.
This issue seems to be resolved in later rtm-proposed releases (I am now
on mako #71). My battery drain now mostly varies between 4-5 up to 8-9
per cent an hour with light to moderate use.
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is
** Changed in: ubuntu-ui-toolkit (Ubuntu)
Status: New = Invalid
** Package changed: ubuntu-ui-toolkit (Ubuntu) = powerd (Ubuntu)
** Changed in: powerd (Ubuntu)
Status: Invalid = New
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages,
54 matches
Mail list logo
