maemo.org - Talk

maemo.org - Talk (https://talk.maemo.org/index.php)
-   Maemo 5 / Fremantle (https://talk.maemo.org/forumdisplay.php?f=40)
-   -   [Announce] CSSU Testing thread (https://talk.maemo.org/showthread.php?t=80525)

freemangordon 2012-12-31 10:54

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by sixwheeledbeast (Post 1309262)
Is there anything wrong with swappolube?
Personal preferences and card speeds means everybodys setting maybe different.
Well if using SD swap which IMO is a good idea
I don't think you can go far wrong with slight tweaks to the "proposed" swappolube setting IMO.

I too agree wiki is the best place if compiling data like that.

Not saying there is something wrong with swappolube settings, besides they fall in WFM category. And i would like to have more scientific approach as CSSU is aimed to every n900 owner. Thus the "CSSU VM settings" should be carefully balanced based on hard values assessment. Otherwise we will go the crappatch route.

@qwazix - AFAIK the only change made by Nokia is swappiness. It looks like the value of 100 set by them is the result of paradigma "read it from swap, not from the fs". And AFAIK this is the one which plays big role in "Unable to answer the phone" reports, along with cluster size, nr_requests, etc.

don_falcone 2012-12-31 11:37

Re: [Announce] CSSU Testing thread
 
May i suggest reviving this particular thread?

mr_pingu 2012-12-31 11:56

Re: [Announce] CSSU Testing thread
 
Now I experienced some weird rotation-bug. I were lying in bed and writing something in MicroB. For the record, I don't have portrait-mode enabled for MicroB. While keyboard was open 100%, for sure (Keyboard was illuminated). Now I pressed ctrl backspace and the screen rotated and the taskswitcher + desktop was portrait (while keyboard was slide out). :S


Oh and Happy new year ;)

sixwheeledbeast 2012-12-31 12:28

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by mr_pingu (Post 1309393)
Now I experienced some weird rotation-bug. I were lying in bed and writing something in MicroB. For the record, I don't have portrait-mode enabled for MicroB. While keyboard was open 100%, for sure (Keyboard was illuminated). Now I pressed ctrl backspace and the screen rotated and the taskswitcher + desktop was portrait (while keyboard was slide out). :S

It may not be just testing.
I have had the desktop in portrait with the keyboard open since updating to Stable 5. Closing the keyboard goes back landscape.

After rotating the device about it fixes itself.
Only happened twice but I can never reproduce, which is annoying.

FIQ 2012-12-31 15:17

Re: [Announce] CSSU Testing thread
 
When HAM does a core update (those which close everything and prevent you from doing everything), it closes every application, *then* check if battery is high enough. Order of these should be swapped.

apt has a localization issue regarding confirmations (in the Swedish locale at least). It outputs everything in english, but it expects input in swedish, which lead to awkaward situations (not an exact quote, but this is what happens):

Do this? [Y/n] y
Aborted.

Basically it expects me to use "j" ("ja" which is "yes" in english) instead of "y". Either confirmation output should be changed to J/n (or j/N) or "y" should be the expected answer. (A similar problem doesn't happen with no, as "no" is "nej", i.e. same letter)

Also, minor error in first post but latest release is 7, not 5.

biketool 2012-12-31 16:33

Re: [Announce] CSSU Testing thread
 
I took the advice of this thread and implimented the swap settings, no more bog to death issues. http://talk.maemo.org/showthread.php?t=69973
According to user nightfire there the Nokia settings are regular linux settings optomized for a spinning disc, I dont know enough about swap access to comment beyond what he writes.
My concern and why it I believe should be a canidate for fix in CSSU is that the average Meamo user shouldnt need to install an app to keep his/her phone from rebooting when the watchdog barks just from moving large files from their flash drive, microSD, or over the network.
Perhaps if a hard solution is not implimented swap could be adjusted only if a bogged situation is detected?

enne30 2012-12-31 17:46

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by FIQ (Post 1309455)
When HAM does a core update (those which close everything and prevent you from doing everything), it closes every application, *then* check if battery is high enough. Order of these should be swapped.

apt has a localization issue regarding confirmations (in the Swedish locale at least). It outputs everything in english, but it expects input in swedish, which lead to awkaward situations (not an exact quote, but this is what happens):

Do this? [Y/n] y
Aborted.

Basically it expects me to use "j" ("ja" which is "yes" in english) instead of "y". Either confirmation output should be changed to J/n (or j/N) or "y" should be the expected answer. (A similar problem doesn't happen with no, as "no" is "nej", i.e. same letter)

Also, minor error in first post but latest release is 7, not 5.

it happens the same in italian locale ...

seanmcken 2013-01-02 15:54

Re: [Announce] CSSU Testing thread
 
im having strange problem, when i connect charger to phone it charges for few minutes then automatically shows "Disconnect charger from power supply to save energy" then again after few seconds it shows "Charging" and like this it go on. plz help

GameboyRMH 2013-01-02 16:00

Re: [Announce] CSSU Testing thread
 
Sounds like you're having the dreaded USB port breakage problem:

http://talk.maemo.org/showthread.php?t=37107

hxka 2013-01-02 16:19

Re: [Announce] CSSU Testing thread
 
It could be cable breakage, not port itself.

seanmcken 2013-01-02 16:55

Re: [Announce] CSSU Testing thread
 
cable and port is fine

seanmcken 2013-01-02 17:02

Re: [Announce] CSSU Testing thread
 
after attaching cable it did shake the cable but it did not disconnect but later aftr few minutes automatically it shows disconnected

Tiran 2013-01-02 17:12

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by seanmcken (Post 1310175)
after attaching cable it did shake the cable but it did not disconnect but later aftr few minutes automatically it shows disconnected

@seanmcken, try another charger. Probably in yours charger electronic components are spoilt!

seanmcken 2013-01-02 17:15

Re: [Announce] CSSU Testing thread
 
i tried everything

Wikiwide 2013-01-05 18:42

Re: [Announce] CSSU Testing thread
 
1 Attachment(s)
Quick reply... Before I forget to ask around here.
Since latest CSSU update (to version 21.2011.38-1Tmaemo7.1), my battery discharges very quickly, lasting not more than one day, even in offline mode. I see nothing suspicious with the background processes. Do you?
Top output:
http://pastebin.com/qNvShf6K
Best wishes.
_________________
Per aspera ad astra...

mr_pingu 2013-01-05 19:00

Re: [Announce] CSSU Testing thread
 
Powertop is a must. Please install powertop (extras-devel, sadly) Run powertop with locked screen and post output, much more useful than batteryGraph /batteryeye. These tools run and monitor all the time and are bad for your battery, too. Only use them when you really need a nice graph.

Also, I am not sure if we can blame testing. I am on thumb (is basically testing) and I dont have problems. You might want to open a new thread

don_falcone 2013-01-05 19:25

Re: [Announce] CSSU Testing thread
 
Powertop didn't show anything useful for me too, pertaining same issue as Wiki. Also, after displaying C states, it segfaults.

EDIT: i'm currently testing if this relates to using WL1251 bleeding edge drivers. But i'm not that much thinking it will be...

sixwheeledbeast 2013-01-05 20:24

Re: [Announce] CSSU Testing thread
 
There have been a number of people reporting this issue now.

Something from the last update?

Code:

  * alarm-ui: rewritten, fixed memleaks
  * glibc: rebuilt to add support for ppoll/pselect (needs kernel)
  * hildon-desktop / libmatchbox2: improved white and blacklist, fixed #12177, #12576, #12660, regressions from the last release and portrait dialog windows
  * ke-recv: added dhcp only option for pc-suite usbnetwork, removed autostart of filemanager
  * powerkeymenu: rewritten, fixed memleaks
  * profile-status-menu-plugin: rewritten, shows all profiles
  * rtcom-messaging-ui-portait: portrait css for messaging-ui (might need to reapply usermodifications)
  * upstart: fixed boot fail if lowmem maemo kernel module is not loaded


Wikiwide 2013-01-06 03:23

Re: [Announce] CSSU Testing thread
 
Quick reply...
I cannot make sense of powertop, so here some memorable lines of it:
dss (100% ON), core (100% ON), neon (14% INA, 7% RET, 1% ON), mpu (14% INA, 7% RET, 1% ON), per (14% INA, 1% ON);
IRQ#56 activity 174 i2c_omap;
frequency 250MHz 93.7%, 600 MHz 5.4%, 500MHz 0.9%;
total wakeups 1062, of which 604 timers and 458 IRQ
I could give the whole output if there was possibility it would help, or run-install some other commands. Or I could calibrate the battery... never calibrated it before.
Best wishes.
_________________
Per aspera ad astra...

sixwheeledbeast 2013-01-06 08:43

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by Wikiwide (Post 1311359)
I could give the whole output if there was possibility it would help

Full output is better then all values can be compared.

Total wakeups seems high for idle and screen off.

sixwheeledbeast 2013-01-06 12:36

Re: [Announce] CSSU Testing thread
 
Also worth pointing out it has been reported on KP thread.
So this may not be related to CSSU at all and more KP52.

pusak gaoq 2013-01-06 12:39

Re: [Announce] CSSU Testing thread
 
Hmmm...I got a major problem on my phone....last 5 days, my devices status bar disappear....i cannot enter any app/settings...if i try to open any of them, my phone stuck....

screenshot....

http://img32.imageshack.us/img32/908...3010620270.jpg

can anyone help me please????

hxka 2013-01-06 12:44

Re: [Announce] CSSU Testing thread
 
That happens when hildon-desktop crashes so often that dkms refuses to restart it.

(Maybe it just disgusted by this windows 7 wallpaper)

pusak gaoq 2013-01-06 12:52

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by hxka (Post 1311489)
That happens when hildon-desktop crashes so often that dkms refuses to restart it.

(Maybe it just disgusted by this windows 7 wallpaper)

any solution for it????

(can't change wallpaper until my devices problem is fixed :D )

Wikiwide 2013-01-06 13:04

Re: [Announce] CSSU Testing thread
 
1 Attachment(s)
Quote:

Originally Posted by sixwheeledbeast (Post 1311408)
Full output is better then all values can be compared.

Total wakeups seems high for idle and screen off.

Quick reply...
Here, two outputs from calling powertop, separated by several hours. I should note, just in case, that powertop is being run as non-root from X-Terminal, and no windows are open, no strenuous activities are undertaken - at most, cellular GSM connection, no Internet browsing, no phone calls, no SMS, no GPS. And yes, the screen is locked-black as much as possible: 11 seconds of sleep are ample time to lock the screen with the hardware slider.
Best wishes. Thank you.

peterleinchen 2013-01-06 15:00

Re: [Announce] CSSU Testing thread
 
Reason why your battery drains is obvious: your device always stays in C0 CPU state ("full awake"). According to your usage info, it should be >90% in C4 state...

But why I do not have an idea. Powertop output does not offer (at least to me) any info.

Did you check with simple top, if there is something running in background keeping the CPU up all the time?


--edit
you should run powertop always as root.

mr_pingu 2013-01-06 15:49

Re: [Announce] CSSU Testing thread
 
@peterleinchen, Why does powertop not offer any info? Firstly it offers the C# states, if your phone wont go in c4, you know something is not right and you can look further.

What process or piece of hardware is taking up your batteryjuice, high activity numbers equals high juice-consumption. One time xbindkeys went mad on my device and I was able to find xbindkeys in the list of powertop. Killed it, ran powertop, fixed it.

don_falcone 2013-01-06 16:06

Re: [Announce] CSSU Testing thread
 
Maybe that particular dictionary bug hits? My device at least is in C3 and C4 states often enough.

peterleinchen 2013-01-06 16:15

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by mr_pingu (Post 1311557)
@peterleinchen, Why does powertop not offer any info?
...

I was misunderstood :( Or i was incomprehensible ;)
Of course powertop gives useful info, but not enough to know what is going on in that particular case. Just to know CPU stays in C0 state.

Wikiwide 2013-01-07 06:10

Re: [Announce] CSSU Testing thread
 
1 Attachment(s)
Quick reply...
I still use batterygraph, since it's the easiest way to get a graph of battery's charge. And bizarrely, it showed "charge up" (from 480 to 730 mA) when no charging was done. Maybe, the insane temperature is at fault; between 37 and 49 Celsius, from battery's standpoint; 38 Celsius is the highest today, according to weather forecast.
I will wait and see. Maybe, the bug will have cured itself soon. I had 'red battery' this morning; maybe, the fever has broken today? Ah, my endless anthropomorphism...
Best wishes.
_________________
Per aspera ad astra...
[Edit]
Yes, the current seems to be as low as 6 mA now (is it believable?). Hooray. Don't know exactly how and why this problem resolved itself, though.
[/Edit]

merlin1991 2013-01-07 11:27

Re: [Announce] CSSU Testing thread
 
A new update has hit the repo (21.2011.38-1Tmaemo7.2)

It's the fix for http://googleonlinesecurity.blogspot...-security.html

The update only updates maemo-security-certman, it adds the 2 wrongly issued intermediate certificates to the blacklist.

hxka 2013-01-13 10:33

Re: [Announce] CSSU Testing thread
 
The bug in osso-xterm: there is no control for input for amount of scrollback lines. I could enter something like 9999999999, which will cause xterm to crash immediately and every time it is launched, or negative values.

pali 2013-01-13 11:26

Re: [Announce] CSSU Testing thread
 
@hxka, see this: http://talk.maemo.org/showthread.php?t=64729

hxka 2013-01-13 11:34

Re: [Announce] CSSU Testing thread
 
I know how to fix it. I just thougt it would be better if value will be limited to something like 0-1000000 to prevent this. Report is based on a an actual case of unaware person screwed his terminal this way.

pali 2013-01-13 11:38

Re: [Announce] CSSU Testing thread
 
Should be fixed now:
https://gitorious.org/community-ssu/...210e6e603792a6
https://gitorious.org/community-ssu/...6fd00ea8f3bde7

hxka 2013-01-13 13:50

Re: [Announce] CSSU Testing thread
 
Great, but i was actually talking about upper limit, negative values are pretty safe, they just act as 0: no scrollback.

sixwheeledbeast 2013-01-19 15:37

Re: [Announce] CSSU Testing thread
 
1 Attachment(s)
Quote:

Originally Posted by sixwheeledbeast (Post 1309402)
It may not be just testing.
I have had the desktop in portrait with the keyboard open since updating to Stable 5. Closing the keyboard goes back landscape.

After rotating the device about it fixes itself.
Only happened twice but I can never reproduce, which is annoying.

Finally got a shot of it.
It's not my imagination honest :)

Smaemo6

hxka 2013-01-19 17:02

Re: [Announce] CSSU Testing thread
 
The bug in… don't know what is responsible for locking, systemui?
1. Lock N900.
2. Press power button.
3. Open the keyboard, while lockscreen is present.
4. Wait until screen will turn off.
5. Close keyboard.
6. Press power button.
Lockscreen will appear, but will immediately disappear, leaving unlocked device. Power button menu will not work until you “really” unlock device with keyboard or slider.

panjgoori 2013-01-19 17:02

Re: [Announce] CSSU Testing thread
 
there is bug in Calculator app. When you open calculator app and do the first calculation it displays result wrong. And when you calculate the same thing second time it gives the right answer. screenshot attached.

http://i.imgur.com/mgHaSkD.png

freemangordon 2013-01-19 17:16

Re: [Announce] CSSU Testing thread
 
Quote:

Originally Posted by hxka (Post 1315921)
The bug in… don't know what is responsible for locking, systemui?
1. Lock N900.
2. Press power button.
3. Open the keyboard, while lockscreen is present.
4. Wait until screen will turn off.
5. Close keyboard.
6. Press power button.
Lockscreen will appear, but will immediately disappear, leaving unlocked device. Power button menu will not work until you “really” unlock device with keyboard or slider.

Can't reproduce here, osso-systemui-tklock - 0.3.1.0+0cssu0


All times are GMT. The time now is 06:49.

vBulletin® Version 3.8.8