[Bug 307374] New: kpowersave: random charge time
https://bugzilla.novell.com/show_bug.cgi?id=307374 Summary: kpowersave: random charge time Product: openSUSE 10.2 Version: Final Platform: i586 OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Mobile Devices AssignedTo: behlert@novell.com ReportedBy: bluedzins@wp.pl QAContact: qa@suse.de Found By: --- I unplugged my laptop so it ran on batteries, kpowersave showed 10 minutes to full charge, then over an hour, then 20 minutes, then close to 2 hours, then 10 minutes. It is completely unreliable estimation. I made two screenshots if anyone is interested. [ I have a feeling I posted such report before but I cannot find it now, maybe it is in opensuse 10.0 section which is closed for me now. So sorry in advance if it is a duplicate, especially of my own report :-) ] -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=307374
Christian Zoz
https://bugzilla.novell.com/show_bug.cgi?id=307374#c1
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c2
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374#c3
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c4
Maciej Pilichowski
since this looks like a problem with your machine or battery.
Does not look for me that way :-). There are too many suspicious things. For example -- I unplug my computer, KPS shows that computers run on batteries (tooltip), icon is changed, but for several seconds after it still shows that batteries are charging (tooltip). It does not make any sense, yet KPS shows such info.
And attach /tmp/batterylog as tar.bz2 to the bug if you could reporduce the problem.
Yes, without a problem. What I spotted so far: * after unplugging there is awful long time with 0:00h to fully recharge * when the CPU changes speed power consumption changes, but oddly, 2GHz takes more than 800MHz * time of recharging does not scale accordingly to CPU speed change -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=307374#c5
--- Comment #5 from Danny Kukawka
Does not look for me that way :-). There are too many suspicious things. For example -- I unplug my computer, KPS shows that computers run on batteries (tooltip), icon is changed, but for several seconds after it still shows that batteries are charging (tooltip). It does not make any sense, yet KPS shows such info.
That depends on HAL (which depends on your hardware/battery information), there is no way for KPowersave to force this.
Yes, without a problem. What I spotted so far: * after unplugging there is awful long time with 0:00h to fully recharge
same as above
* when the CPU changes speed power consumption changes, but oddly, 2GHz takes more than 800MHz
Sorry, but I don't understand what you mean
* time of recharging does not scale accordingly to CPU speed change
Why should it? This depends all on the info of your battery and the elapsed time -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=307374#c6
--- Comment #6 from Maciej Pilichowski
Does not look for me that way :-). There are too many suspicious things. For example -- I unplug my computer, KPS shows that computers run on batteries (tooltip), icon is changed, but for several seconds after it still shows that batteries are charging (tooltip). It does not make any sense, yet KPS shows such info.
That depends on HAL (which depends on your hardware/battery information), there is no way for KPowersave to force this.
Danny, I won't believe HAL displays tooltips for KPS, come on :-)) KPS should have some common sense and not display incorrect information. Ok, those information could come from faulty hardware or faulty subprocess, but they are easy to correct. After all computer cannot be unplugged and plugged at the same time, right?
* when the CPU changes speed power consumption changes, but oddly, 2GHz takes more than 800MHz
Sorry, but I don't understand what you mean
The power consumption should be lower when CPU is running at 0.8GHz and higher when it is running at 2GHz.
* time of recharging does not scale accordingly to CPU speed change Why should it? This depends all on the info of your battery and the elapsed time
Because power consumption varies, right? Run computer at lower speed it will last longer, set everything to high values (brightness, speed, etc.) and battery will not survive for long. Thus time should be scaled according to power consumption.
-- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=307374#c7
--- Comment #7 from Danny Kukawka
Danny, I won't believe HAL displays tooltips for KPS, come on [...] No, KPowersave display/generate all it's own tooltips, but all the infos are from HAL. And we decided a long time ago that we trust HAL to get all info, KPowersave should not contain code to workaround broken hardware. Everything depends on the hardware. And in interest of power consumtion HAL don't poll every second for new battery information (if the battery don't send events if something change). Because of this it can take some time until the battery info is updated.
The power consumption should be lower when CPU is running at 0.8GHz and higher when it is running at 2GHz. Maybe it is lower (I would assume it), but (again) all these info depends on HAL. You should not assume the value for power consumtion in the KPowersave info dialog is always up-to-date, it show may not every change. It should be updated at least every 30 seconds or on battery events. This should be enough, every other behavior would only cost power of your battery.
because power consumption varies, right? Run computer at lower speed it will last longer, set everything to high values (brightness, speed, etc.) and battery will not survive for long. Thus time should be scaled according to power consumption.
HAL can only calculate the remaining battery time from the elapsed time and your battery info (as e.g. last full capacity, remaining capacity, present rate) and not from the CPU freq. because of this it need some time to get a reliable value for the remaining time and due to this changes (maybe also only spikes) in the CPUfreq don't get directly into the remaining time. You can believe us: this all isn't that trivial as you may think. (and also the Windows estimations for remaining time are not that reliable) ;-) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=307374#c8
--- Comment #8 from Maciej Pilichowski
And we decided a long time ago that we trust HAL to get all info,
I say it as a developer, and not only for KPS but in general -- I disagree. What counts is not only OOP nice code, or nice indentation, or anything like this but outcome -- total user experience. Design in manner "I am just a messenger" does not solve anything. (Please read twice the last part). It is just searching for app/person to blame for fault (strangely enough this is how work in Poland looks like :-))) ). Following this design you get not stable, not reliable software -- who cares then who made mistake. In TUE software (without giving names) failed. Patient died, airplane crashed, GPS lost the track, data are missing, mail wasn't delivered. So, again, nothing personal, nothing against KPS, or something like this, but I am simply afraid of such decision. One more remark -- now about opensuse image. Do you (as observer) really think that user thinks in terms of interrupts, signals, HAL, etc.? I doubt it. For him/her there are windows, suse, macos. That's it -- and opensuse (sic!) gives false readings. Nobody has time to investigate that it is not opensuse, that it is not KPS, that is not HAL, but this is... wait a second, this user already switched to Mac. First impression matters. So either (in general) a product delivers excellent quality or vanishes. Ok, maybe I am a bit of mood for writing, excuse me :-DD but I tested 10.3 a bit and I am so happy with all improvements done, that I am personally sorry every time I see such risky decision (from suse image point of view).
KPowersave should not contain code to workaround broken hardware. Everything depends on the hardware.
Btw. the same decision in cups results that users cannot upgrade to newer version, because newer version is better but doesn't work :-) As developer I rather say, give me hardware and I make it fly (no, not out the window :-D).
Maybe it is lower (I would assume it), but (again) all these info depends on HAL. You should not assume the value for power consumtion in the KPowersave info dialog is always up-to-date, it show may not every change. It should be updated at least every 30 seconds or on battery events. This should be enough, every other behavior would only cost power of your battery.
Sure, but I waited a bit longer, so it is not just a random reading pick.
You can believe us: this all isn't that trivial as you may think. (and also the Windows estimations for remaining time are not that reliable) ;-)
Ah, I am not saying it is trivial or something like that, I am just saying it would be much better KPS would fix what is fixable. Now I use windows only for scanning, so I didn't test it :-D Ok, a bit lengthy comment, but I couldn't resist. Thank you for reading and maybe for reconsidering design concept. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=307374#c9
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c10
--- Comment #10 from Maciej Pilichowski
You miss the point, that we can't workaround every possible brocken battery (ACPI bios or what ever).
I would say it is one of the task of software. Deal with hardware. Broken (I mean in design) or not. Do you remember that story with LG dvd/cdrom drives?
The only thing you can try is (only for your machine): set via a FDI file for all your ACPI battery objects: battery.remaining_time.calculate_per_time=true (bool)
Could you please help me with that? I didn't anything useful, the closest match was file: /usr/share/hal/fdi/policy/10osvendor/10-power-mgmt-policy.fdi but there was no battery entry there. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=307374#c11
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c12
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374#c13
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c14
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374#c15
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374#c16
--- Comment #16 from Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374#c17
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c18
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374#c19
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c20
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374#c21
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374#c22
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User dkukawka@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c23
--- Comment #23 from Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374
User dkukawka@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c24
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374
User bluedzins@wp.pl added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c25
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User dkukawka@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c26
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374
User bluedzins@wp.pl added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c27
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User bluedzins@wp.pl added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c28
--- Comment #28 from Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User bluedzins@wp.pl added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c30
--- Comment #30 from Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User bluedzins@wp.pl added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c29
--- Comment #29 from Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User dkukawka@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c31
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374
User bluedzins@wp.pl added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c32
--- Comment #32 from Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User bluedzins@wp.pl added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c33
Maciej Pilichowski
https://bugzilla.novell.com/show_bug.cgi?id=307374
User dkukawka@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c34
Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374
User dkukawka@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c35
--- Comment #35 from Danny Kukawka
https://bugzilla.novell.com/show_bug.cgi?id=307374
User dkukawka@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=307374#c36
Danny Kukawka
participants (1)
-
bugzilla_noreply@novell.com