Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Stdout can be redirected with '>' to write to a file.

# ls -l > ls-l.txt (creates/overwrites file and redirects output to file)

# ls -l >> ls-l.txt (creates file if not present or appends to existing file)

http://tldp.org/HOWTO/Bash-Prog-Intro-HOWTO-3.html
http://www.mathinfo.u-picardie.fr/as...direction.html

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

:p

//Edit

No, seriously, sorry, but it's quite hard to comment on. Maybe install water cooling system to your charger? :P

Also, remember that it's -devel. Considering what You've just written, I think your technical knowledge isn't entirely up to using -devel (no offense, just advice).

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Pali, I've checked charging with bq2415x_charger. It works great, although, when I wanted to alter sysfs entries to charge with 1250 mA current (I have 3Ah battery) as I used to do with charge21.sh script by ShadowJK, I discovered strange thing. My sysfs entries looks as follows:

Are they supposed to be divided/multiplied by something, to get real mA values? Or are they calculated like that, because my reported capacity of battery is now 2616 mAh (I haven't yet calibrated it properly, since last repair operations on N900)?

Also, charging chip accept 1250 as max charging current, and it seems to me, that with such values, it just uses default 950.

Another strange thing is what happens, when I try to mess a little with those values. If I set /sys/class/power_supply/bq24150-0/charge_current to any small number (for example, 300), it changes itself to 1350 (!). real charging current becomes something about 500 mA, then. When I try to set my desired charging current, 1250 (mA), it changes itself to 2050.
---

Could You, please confirm/deny that it's bug, and if the latter, explain how it's supposed to work?

2nd question is purely cosmetic - during charging, diode stays yellow/bright orange, like in emergency charging, even with screen ON. You've mentioned, on IRC, that it's possible to disable it via sysfs entry - which one?

Also, you've postponed element, that will use charging pattern boost from mce.ini (just like usbmode.sh does, during booston) for later, yes?

Thanks a lot for answers, in advance.

/Estel

// Edit

I have noticed some tendency, with those sysfs entries. If I set it to 950, it changes (as usual, on it's own) to 1750, like it was by default. When I set 1250 - as said before - it changes to 2050 = 300 more than 1750, just like 1250 is 300 more than 950.

Could it be, that it works OK, but just need to be "translated" somehow? In such case, 1350 would translate as 550, i.e. lowest charging current possible by chip (outside saturation charge, towards end), so that is why it changes itself to 1350 as lowest one?

If yes, how to translate termination current, that present even more strange results? Or current_limit, which sets itself as 1800 max? It would be nice, if You could explain how all values - which I've listed in first part of this post - should be calculated.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

What is advantage in charging with bq2415x over ordinary charging with bme and how to activate it exactly?


p.s. Sorry if the question is stupid, I am a noob. ;)

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Another thing spotted - mentioned earlier:

Acts very strangely. It is supposed to mean 100 mA termination current, as lowest value it accept is 6500, which is 50 mA, and lowest value accepted by chip. Hoever, highest value accepted by chip is 400 mA, which should be 6850 as sysfs value. Unfortunately, it accepts 6800 as max (changes itself to 6800, if higher value is presented)

Which is even more confusing, after changing it, You're absolutely enable to provide default value (6550) again - whatever one input there, it changes itself to 6500 or 6800, *only*. It refuses to accept anything in between, as said, even value that was set by default. fortunately, unloading module, enabling bme, disabling bme and loading module again, bring all related sysfs values to default.

I may be wrong, but this one sounds absolutely like bug to me.

/Estel

// Edit

branja, to charge with bq2415x_charger, You need to, as root:

To stop using it, You do:

Advantages - it's huge step toward introducing replacement for bme. Currently, only two direct advantages for end-user apply - possibility to set own charging current (higher one for bigger batteries, for example) and termination current (for terminating charging properly during heavy load), and possibility of charging during USB hostmode.

Basic goal is to replace BME, but of course, being totally open, bq2415x_charger have virtually countless possibilities. It should be easy to write scripts/program for - as example - enabling, when invoked, charging only to 80% (save battery health), or to introduce leaving device on charger without negative impact on battery (now BME doesn't necessary follow best practices re extending life of battery).

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Hello.
How do I set my own charging current?
I got the new BL-5J with the additional 100mA and there for the bettery only get charged to 85%.
See this thread:
http://talk.maemo.org/showthread.php...ighlight=bl-5j

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

I'm fine with using devel since 3 years and never had a problem. Also this time my problem gone after downgrade.
After using n900 for 3 years my charger never has been so hot. I don't need a water cooling, just stable kernel(as stated in the topic).

I post this information to warn other users cause this may damage the hardware. With all respect for your work on developing the kernel, there is no need to flame people who are describing symptoms of bugs you made.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Hell

When i install KP51 and switch off the device than press on but nothing ! My n900 is dead and now do not open

When i press power button and appear nokia ikon on the screen short time than vibrate once and device goes turn off than repeat again this intervals. But don't open any time !

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

I think I found a bug

stop bme
modprobe bq2415x_charger

->Display turns off while charging: I wanted to leave it on because sometimes it cannot be unlocked, so I have to remove the battery.


//Edit: Okey with simple brightness widget the display can run 24/7
I noticed that SSH works even if the display is off and not unlockable. Maybe this can be used to get debug information.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

i remove the battery but same result device is do not open. When i connect to charge device is open itself than appear Nokia ikon than device goes off than on intervals.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

i think one way is format to my n900 ! Thx KP51

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

One word - bullsh|t.

kp doesn't have anything to do with Your EXTERNAL charger temperature. Theoretically, it could get warmer, if You would use bq2145x_charger module with manually set higher charging current, but even this is unlikely to happen.

Proper "bug report" would contain any useful data - like monitor of charging current, when your charger becomes hot (that would prove or deny something fishy going out), etc. I don't care if You're using -devel for 3 or 300 years, both content of what You've written, and way in which You have written it, mark You as person that should stay away from -devel, in my humble opinion (and You have right to not agree :) )

/Estel

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Pali, as per:
http://mg.pov.lt/maemo-irclog/%23mae...08-02T01:19:39

...Joerg doesn't have slightest idea why bq2415x_charger is acting like I've described in last page. Considering, that You followed his formula, it isn't expected result.

Could You compare things I've reported with what You wanted to achieve, and check for bugs in code?

/Estel

// Edit

Also see from here:
http://mg.pov.lt/maemo-irclog/%23mae...08-02T01:35:23

...to the end of discussion. wrong RS value is candidate to being responsible for insane values, but, still it doesn't explain fubar'ish behavior of termination_current (described in my earlier posts). Something is definitely wrong here.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

As I use bq2415x_charger exclusively for every charging, I've noticed another 2 glitches in bq2415x_charger:

1. Using default settings, charging doesn't terminate properly - it terminates *too early*, i.e. before battery reach full capacity, and before vdq is turned to 1. This prevent calibration of battery sensor, and usage of our last 5% of battery capacity ;)

2. To overcome point 1, I've investigated buggy behavior of termination_current sysfs entry, mentioned earlier. I discovered why it accept 6500 and 6800 only, as value - just like charge_current it's strange translation of real current, but unlike charge_current, it accepts *only* real values.

For example, with charge_current, to have 950 mA current, You must specify 1750 as sysfs value. to have 1250 mA, You must put 2050 there, and so goes on. Values are accepted, and remain this way.

For termination_current, hoever, to have 50 mA termination current, You must specify exactly this - 50. It gets "translated" to 6450 then, and sysfs entry turns itself to 6450 (which means 50 mA). If You want 100 mA, You need to specify 100, and it translates itself to 6500 - etc, up to 400 of termination current (which translates to 6800).
---

It's definitely a bug (or chain of bugs), but at least it works (when You RE it's twisted logic) and allow full control over charging current and termination current.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Ok
i try to upgrade kernel (kp51) two days ago and crashed my system. But today, i try again to install kernel (kp51) but this time, it install perfectly. Now working very well ! Cleven is ok, and o have not any charging problem ! i think new kernel is good job.

PALI thank you very much ! Sorry my other posts.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Hi Estel,

there is sysfs entry registers which output raw value of each bq24150a register. Look at datasheet http://www.ti.com/lit/ds/symlink/bq24150a.pdf (page 29) to check if all values are correct. Also you can ask Doc about raw register values.

registers sysfs entry is read-write. You can set to register 0x01 value 0xFF by: echo "0x01 0xFF" > registers

EN_STAT is for enabling/disabling (emergency) charge led.

Then rmmod bq2415x_charger driver and use charge21.sh script for charging... Use i2cget to raw register values and check if values are correct.

Next week I will be without internet, so I cannot look at it.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Pali, I've looked at it briefly, and raw values via i2cget (charge21.sh) looks OK. It seems that they're wrong only when bq2415x_charger is used.

I'll look at it later, as I'm also starting 2-weeks vacations :)

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

my charger is AC-10E i haven't any problem

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Well, point taken. I guess you are right, this inconsistency should be solved somehow (either moving some stuff to e.g. /opt/additional-kernel-modules and symlinking them as e.g. /lib/modules/2.6.28.10-power51/extra/* or placing them all in /lib/modules together... probably the first option is a bit better, as if someone needs some special modules in rootfs to be loaded at boot, she/he will have sufficient knowledge to copy them by their own to rootfs manually).

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

:) lol, i found a workaround for the non unlockable display. I don't lock it and start bedside instead (doesn't really work but turns my screen black). To "unlock" I press strg shift x so bedside is minimized and I can close it again.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

okay I have mild bugd related to batterry/charging. I couldn't describe it with details, but I may collect any logs you want (I have full acccess to root account on demand as I suppose most people here, but it is not bad to remind). On medium usage and everyday charging temperature is mostly around 40 celsius. My medium usage includes some usage modem mode (via qtMobileHotspot) some fm transmitter usage and mostly 2G network (display show 2.5g). I've been sure I now place my N900 on right of my laptop (on left is mentioned hot air flow). Temperatures are measured on batterry via bnf on single demand measure (no monitoring) and probably through i2c mechanism (idk how bnf work ExactlY). Extremes measured were 32C on colder day and 52C on high hardware usage plus charging but is mostly stable around 40C - and it's summer here (I say it once I realised only quater of earth have summer now and if I were on zero deegrees north - i couldn't remember the earth line name in english - it would be a special case because it would be even hotter than here - temperatures of atmospheric air are mostly below 35C here).

I have some problems with other batterry metters (mostly default nokia one it got confused about the real batterry temporal capacity a few times but else it showed properly). I do not seem to have instabilities separate from sometimes too big temperature to my flavour (but I haven't properly observed my device before since I haven't batterry i2c script and bq module for batterry was default blacklisted and I haven't touched my modules list).

rest would be later.

PS: my device is mostly on 500mhz so i checked the overclocking or always-100-percent-processor-programs are not the issue.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

What's supposed to happen when the battery is full?
I have been charging 24h with display on, but the battery icon still shows the charging animation and the yellow Led is on.

Also I have to unlock my sdcard after every reboot to get write permission: chown -R user:users media/mmc1/

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

hi

here is my problem : i cannot connect my N900 to the internet but i can download packages manually to my pc . Is there a way to install kernel power through terminal ? And the magnet sensor of keyboard is always giving a closed state .
Sorry if it has been asked before

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

@myname24,

download the .deb's directly (go to maemo.org/packages, type kernel-power, and download relevant packages) -- SEE FIRST POST for instructions on how to install them.

Add.: @myname24, please respond here, not per PM. Post the exact output of the dpkg commands you execute. Otherwise nobody can help you.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

well it didn't work and give me conflicting package with maemo kernel and pre dependecy errors

edit :
Nokia-N900-51-1:~# cd /home/user/MyDocs/kernel/
Nokia-N900-51-1:/home/user/MyDocs/kernel# dpkg -i kernel-powe
r_2.6.28-10power51_armel.deb
dpkg: regarding kernel-power_2.6.28-10power51_armel.deb containing kernel-power:
kernel-power conflicts with kernel-maemo
kernel-maemo (version 2.6.28-maemo19) is present and installed.
dpkg: error processing kernel-power_2.6.28-10power51_armel.deb (--install):
conflicting packages - not installing kernel-power
Errors were encountered while processing:
kernel-power_2.6.28-10power51_armel.deb

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

@myname24,

Well, then remove kernel-maemo first. It's now obsoleted by kernel-power.

If you also have kernel-flasher-maemo, remove it as well.

Then try again.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

edit : now works thanks

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

See my last few posts about termination_current, and sysfs entries et all.

After another few charging cycles, I've noticed that neither default 150 mA (6550 as sysfs value, when read) nor 50 mA (6500 sysfs) works OK - first one terminates charge too early, without switching VDQ to 1, while latter never terminates it (in case of my dual-cell battery, result for regular ones may wary).

6500 (100 mA) seems to work best - IIRC, it was also default used by BME?
---

As for SD card, no such problems here.

/Estel

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Can some1help me with his pls.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Try to remove the backports-modules package, then retry to update the kernel.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Sorry for not posting "the rest" but I forgot what it was. Today I had batterry/charging bug I suppose. I couldn't boot my device for long time after crash or emergency shutdown. It succeeded without reflash after long time of retries. The reason why I think it is related with batterry state is the batterry was almost empty at the shutdown or another problem forced my device to shutdown.

I want also to remind old known bug I suppose - in kernel power with multiboot implementation (read it as - not direct booting only kernel otherwise) it fails to boot properly into system while charger is connected - it seems to lock at ACT-DEAD state despite trying to leave it in a way which worked previously. I don't know if it is the same bug as one from ages but rarely occuring or it was new since Kp51 - but I feel forced to remind that - even if user resolution/workaround is trivial - just to disconnect charger, boot up until desktop shown and reconnect it. I hope you would find some time to adress that.

PS: the time it failed to boot was too long for the time it having too low batterry to do the manuever, and I also had the fbcon log with error which I couldn't catch long enough to know what were written, but long enough to see there was at least one then.

EDIT: it seems reseting the device by pulling out batterry for 30 seconds were neccessary to recover non-booting problem but not enough anyway.

---------------------------------------------------------------

it seems from the behaviour that kp51 after "full" batterry and wallcharger present tries to overcharge the batterry by mistake but it may be wrong device Charging Reporting behaviour under kp51.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

It did work.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

What am I doing wrong?
I'm trying to load the injection drivers but it dosen't work!

and its giving me error that My kernel version does not support injection
Or sometimes its just go one line down and dosen't write nothing.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

Today I had Kernel Crash. After it it rejected trying to power up. Wise by previous experience I removed the battery (and put in after 30 seconds) at it booted properly at first time after reset. After booting it showed info about kernel crash (blah blah not properly blah blah custom settings reseted) and message looking like "batterry almost empty". So I checked it by bnf and it was at 33% (nokian batterry meter showed small red stripe) and restored my custom settings ("kernel-config load MAJC89" and "kernel-config default" from root console). Then I restored my date and time over internet by some "ntpd -s". Used it a much for 30 minutes without a problem and it shutdowned itself again (but now because of real batterry discharge). I'm now recharging and boot was clean (booted properly at first time, no settings-reset message, and guess what - it didn't bugged boot up, despite charger beeing present and connected - no wallcharger bug this time at first boot trial). I'll check if any charging bug will occur or if my N900 piece is cured for at least a day (all the kernel-power power-maintenance bugs seems to have some trigger without which, they don't seem to occur at all).

PS: I noticed that during my kernel crash I had mediabox (which is audio/video player) running and playing and headphones on my head. It were playing with stutterring for 3 seconds after screen went off. It may be related because mediabox during switching tracks with shuffle (which lasts between 50ms to 750ms) seems to be on 100% processor which triggers my overclocking settings temporary for the fragment of second to 900 mhz (and after it returns safely to 500 mhz which is my minimum in my kernel-config profile - I checked it with freq-watch.sh which i noted from the quote when I wanted to know why "kernel-config show" show the 900mhz and then it was obvious that kernel-power-config fails to measure the current frequency because it use too much processor resources, so I found the script very helpful for the task that "kernel-config show" fails to do). I remind that I previously late time ago (when latest VDD2 patch came in) I tested it can stand 900mhz for 15 minutes without single problem, so kp51 was partially the reason (either my kernel-config profile is valid or beacame invalid after update to kp51 and my profile didn't changed since then).

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

http://www.debian-administration.org..._in_14_seconds

we can do this on n900?

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

kernel is already optimized for device, it doesn't have to run anywhere else so all required functionality should be built in. wifi drivers aren't because of need to swap with injection drivers.

if your looking for fast booting, your best bet would be to replace the current init system with systemd or maybe upstart. however, with systemd, this would involve reworking all the current init scripts to become .service files to make the most of it.

question is, is it really worth it? device has long battery life compared to laptop. its usually not (re)booted very often as it needs to be always on.

edit: appaerently already uses upstart.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

@Android_808
i just meant that the phone should boot between 7 to 8 seconds. thats it

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

@majaczek-how did you judge you had a kernel crash? It could be that I miss something, but I fail to see your crash backtrace (/dev/mtd2). Could it be that your battery no longer holds the charge and the device has shut down? Also you use some custom OC profile-care to share it?

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

might be limited due to i/o speeds or upstart version lacking some optimization. if an initrd is used it tends to slow it down a bit. can't tell at the moment, rather worryingly /boot appears to be empty, despite kp and k listing files installed there.

my eee 1000 with standard ssd's and a 2gb ram upgrade will easily go from grub to lightdm login screen in 4.8s according to systemd's analyize app. with a slower processor, less ram and probably slower flash memory, i imagine n900 with tuning will still be more than 8.

if it's something your interested in looking into, start another thread as it's a bit off topic here.

Re: [Announce] kernel-power stable v51 in Extras-Devel
 

n900's flash memory and microsd cards are a lot slower i/o-wise than SSDs, and so are the rest of the specs