Re: [Announce] kernel-power v48
 

Is h-e-n working on it, now, while copying large files (200mb+)? On h-e-n thread I was informed it is not a h-e-n problem... Thanks.

Re: [Announce] kernel-power v48
 

Have you been contacted yet? If not, I received a message back on your problem.

The package got rejected by the queue manager because the changes file has issues: kernel-power_2.6.28.orig.tar.gz size mismatch (66766084 instead of 67389527). So you must have uploaded different source package with the same name. Can you please check and try again.

Re: [Announce] kernel-power v48
 

No I have not been contacted yet.

I used source tarball (kernel-power_2.6.28.orig.tar.gz) from kernel.org. And in repository is other version? And this is reason why v47 is not on maemo.org/packages interface?

Re: [Announce] kernel-power v48
 

Afaik you should use source got with apt-get source kernel-power, not one from kernel.org

Re: [Announce] kernel-power v48
 

Did anyone test what effect v48, especially the 'compcache/zram patch from kernel-bfs' (which implements ram compression as far as I understood it) has in daily use compared to v47? I was really hoping for a reduction in swapping activity. And it may be reduced, I am not sure, I did too many changes to reliably tell.
I was secretly hoping Fennec would run better, but that does not seem to be the case.
Or is it something we need to activate manually?

Re: [Announce] kernel-power v48
 

Some usefull info about BQ27200 chip here: http://mg.pov.lt/maemo-irclog/%23mae...07-26.log.html
ShadowJK wrote, how to do learning cycle which should fix LMD:

Re: [Announce] kernel-power v48
 

You wanna check out this bro.

Learning achievement unlocked!

Re: [Announce] kernel-power v48
 

Pali, how do we turn on ramzswap/compcache?
The kernel modules seem to be missing.

Re: [Announce] kernel-power v48
 

You have to wait for tigerite to finish it off. This is just support built into the kernel.

Maybe try paypal'ing him some beer?

Re: [Announce] kernel-power v48
 

I found some source code of TI DSP libraries here: https://gforge.ti.com/gf/download/fr...23i3.8.tar.bz2

There are for example jpegdec, jpegenc, videodec, videoenc which are in fremantle closed: http://maemo.org/packages/view/libomxil-ti-components/

Is TI DSP sources compatible with N900 and Fremantle?

Re: [Announce] kernel-power v48
 

They should be, given that other parts of OS side DSP are replaceable by drop-in.

Re: [Announce] kernel-power v48
 

I can confirm,

flash OS + eMMC, installed kernel-power v48 + settings 0.12--->black screen on boot.

flash OS + eMMC, installed kernel-power v47 + settings 0.11, upgraded to v48 + settings 0.12 ---> successful! (uname -r is showing 2.6.28.10-power48)

Also, MobileHotspot does not work with v48. Says the dependency "wireless-tools" was not installable.

Cheers!
Joshua

Re: [Announce] kernel-power v48
 

This confirm, that there must be bug in preinst/postinst kernel-power scripts. I will try to find it.

Re: [Announce] kernel-power v48
 

Also...
flash OS + eMMC, installed kernel-power v48 (no settings) ---> successful!

Re: [Announce] kernel-power v48
 

settings package install
* overclocking script (+ ability to overclock on boot)
* script which load all modules specified in /etc/modules on boot
* script which can be configured to run fsck or ssh on usb at early boot

Re: [Announce] kernel-power v48
 

So... I'm not a devolper... It's only an idea... Maybe when install kernel-power-flasher some modules aren't written correctly.

Sorry for my bad english

Re: [Announce] kernel-power v48
 

Just to help, one more installation case...

Installed V48 after a long time using V47 (no uboot, no multiboot, no nitdroid, just plain V47) in next order

V48 kernel
V48 modules
V48 flasher
V48 settings

After reboot device hangs on black screen.

Re-flashed eMMC and Kernel, booted, installed V48, same order like above (no settings) rebooted, works ok.

Installed settings, hang on black screen.

Now I am running completely re-flashed device and configured for me, with V47 and scared to try to upgrade...

Re: [Announce] kernel-power v48
 

successfully upgraded from v47 to v48 with kernel power settings 0.12!!

is there any improvement in battery usage time??

Re: [Announce] kernel-power v48
 

*facepalming @ 500-900 mhz (limits) rate*

Do you see any things related to battery usage in changelog?... Do You see *any* things related to battery, in *any* kernel-power version changelog?... Did kernel-power *ever* get anything to do with batteries, except false reports of battery temperature readings?...

Re: [Announce] kernel-power v48
 

im really sorry for asking such stupid question. it is true i didnt saw anything related to battery improvement in any kernel version. but please dont talk to me like that. im having some hardtime with my battery. again sorry for my stupidity.

max freq 600 giving more speed than before. testing the device to see if its stable on v48.

Re: [Announce] kernel-power v48
 

upgraded from v47 to v48 a couple of weeks back . using backup menu and multiboot . everything worked perfect straight away and no problems at all...

Re: [Announce] kernel-power v48
 

Same problem in qemu - After installing settings package black Xorg screen. (Without settings package + manually compiled qemu kflges module - all ok)

But there is no error, no warning in dmesg or serial console...

Re: [Announce] kernel-power v48
 

Problem in qemu was found. But I do not understand why it works in upgrade v47-->v48 and setting11-->settings12 and not with v48+setting12.

In /etc/default/mount-opts is KERNEL-POWER-HOOK line which run script /etc/kernel-power/pre-mount (if exists). But if this script does not exists it hangs too (WHY?) And why it does work after upgrade?

Re: [Announce] kernel-power v48
 

I should correct myself. MobileHotspot does work with power48. I originally did not have the repository containing "wireless-tools" enabled in App-Manager. After enabling the repository, (Fremantle Tools), all is well...

So far, power48 is working well, though the vibration from touching the touchscreen is somewhat prolonged occasionally. This could be my core-voltage settings, though.

Cheers!

Re: [Announce] kernel-power v48
 

Perhaps something to do with an updated version of ash in busybox power?

Re: [Announce] kernel-power v48
 

it is in qemu with clean pr 1.3

but something new I have. Problem (in qemu and I'm sure same on real n900) is that /home and /opt was not mounted, so hildon-desktop cannot be started (black Xorg screen). But why /home is not mounted? /etc/event.d/rcS-late is responsible for mounting. So I enabled console output from this script (then double called sync), but then no error and qmeu normally bootmed me.

So I again created clean qemu image, installed kernel v47 and shutdown. All ok. Started again installed setting12, enabled debug output from rcS-late, double sync, shutdown. Started again, no errors in rcS-late, home mounted and all is OK.

I do not understand why it works if I enabled debug output (changed 'console none' to 'console output' and added some echo...). I ony called $ sync and $ sync.

Re: [Announce] kernel-power v48
 

Finally problem with kernel-power-settings was found and fixed.

Problem:
I'm using madde for compiling and creating deb packages. Madde has its own dpkg tools (own tool for low-level creating deb package) which ignore file permitions (or change it or use its own??). And it still change permitions of /etc/kernel-power/pre-mount from 755 to 644. And upstart failed in /etc/default/mount-opts (which started pre-mount script).

Solution:
Changed generating file /etc/default/mount-opts which execute script from method 'if test -x FILE && ./FILE' to 'sh FILE || true'. In qemu now all works.

Why it worked with update? Config file /etc/kernel-power/pre-mount was not changed, so dpkg does not replaced it when update from 11 to 12. (and permitions was kept).

When I debugged it I also chmod +x permitions, so I was unable to debug it...

New fixed version is on same place.

Re: [Announce] kernel-power v48
 

Curiosity, it could be possible to generate a power kernel with Android support ?

I don't know what patches are applied by Android/Nitdroid's team to kernel but I have wish to try to run Nitdroid on a chroot enviroment (like SuperJumbo for BeagleBoard).

To do so we need a kernel which can boot Maemo and Nitdroid (full boot)

Re: [Announce] kernel-power v48
 

Dear Pali and kernel-dev team

All the work you're doing is great. It's amazing that you're able to maintain the momentum and keep us informed of what's happening. My question is - when the version nudges on to .49 or .48 becomes stable enough to start developing addons, please would you start a new thread for each addon/update e.g "kernel development: abc for v.XY"?

I ask this for a few reasons:
- kernel development extends across so many features thus questions and discussions will be equally diverse. Therefore, they [dicussions] can appear disjointed, which makes finding answers a bit difficult and will in turn result in a lot of people repeating the same question.
- Reading through over a thousand posts to get up to speed on testing things to watch out for is beyond most of us!
- The thread title doesn't accurately reflect its content: it started by pulling together the v0.46 patches, it goes right through all the v0.47 work and now we're discussing pre-extras-devel 0.48.

Consiolidation and evolution are good but but it's quite confusing, especially with the number of sub-topics.

On a completely different note, have you had any news about updating the package interface. I see Titan is still the sole maintainer for a few of the associated packages - might that be something to do with it? A work-around, which would be a pain to start withe because it would break the auto updates would be to upload a package with a new name, say "power-kernel", instead of "kernel-power"? If you did this it might be an idea to make a member of the council a silent 2nd maintainer so it can "change hands" easily in case you can't/dont't want to continue?

I would also like to reiterate that both the code and efforts to update the docs are amazing and really help us all to make the most of this device. I have noticed the irony of this now slightly verbose, multi-themed post so, over-and-out!

Re: [Announce] kernel-power v48
 

After discusion on #maemo with DocScrutinizer and ShadowJK we decided to disable one i2c patch which break API and can damage n900 (patch was added in v45). When this patch is disabled - bme is broken when bq modules is loaded.

So autoloading bq module is disabled too and now is not possible to have both bme and bq. (bq module is still installed so after stopping bme is possible to manually load it).

Now I recompiling v48 and publish new version as soon as.

Re: [Announce] kernel-power v48
 

aww man, I was just getting into the bq27x00 'true' capacity readout.

I ask that joerg please explain what exactly the problem is with bq + bme simutaneously.

No seriously, I know I2C. WTF is the damage?

Re: [Announce] kernel-power v48
 

See: http://mg.pov.lt/maemo-irclog/%23mae...07-28T16:54:58

Re: [Announce] kernel-power v48
 

pali i have been using bme + bq since the day bq was released
and i'm having no problems
and i expect no problems

Re: [Announce] kernel-power v48
 

updated new version

Re: [Announce] kernel-power v48
 

// Edit - scratch that, I reconsidered this and I was wrong here. I'm not deleting it, to not cause context havoc with posts after it

// Edit - scratch that, I reconsidered this and I was wrong here. I'm not deleting it, to not cause context havoc with posts after it

Re: [Announce] kernel-power v48
 

Not really that useful.. but I can "confirm" that BME + BQ has caused no problems for me (with 46 and now 47).

A few days ago during a clean-up I uninstalled AP and blacklisted BQ, because I wanted to simplify things, and I've always been happy with the battery values reported by BME.

But in any case, BQ never caused a problem for me.

Re: [Announce] kernel-power v48
 

I just re-flashed the N900 and I installed kernel-power v48 ... If you want I can do some tests ...

Re: [Announce] kernel-power v48
 

joerg_w here explains the reasoning behind the problem with the I2C patch. While I do understand where he is coming from I am unconvinced that a multi-read or multi write operation being performed by a process would have a 'relinquish to scheduler' in the middle of writing/reading multiple bytes on the I2C bus. Well, I certainly wouldn't but fck knows what happens inside bme, in addition to this joerg_w does know what he is talking about.

How would you identify a confirmed case of I2C collision? A random reeboot here? A funny battery percent for a couple of seconds there? The fact is, there could be collisions occurring more often than we think.

What I would like is for battery charging to be taken over by either bq module (perhaps not the right place) or some replacement for BME. Apart from handling charging and updating HAL with some erroneous data it doesn't really do much else that I can want.

Re: [Announce] kernel-power v48
 

Camera Error I don't know if is the KP or something else
http://talk.maemo.org/showthread.php...56#post1060456

Re: [Announce] kernel-power v48
 

Hi all,

I'm trying to implement Application level filtering.

I found out that it could be achieved with iptables.

So this is a sample rule that should deny acroread program to use network:

iptables -A OUTPUT -m owner --cmd-owner acroread -j DROP


The problems is that it does not work on maemo with KP48.

iptables -A OUTPUT -m owner --cmd-owner acroread -j DROP
iptables v1.4.6: unknown option `--cmd-owner'
Try `iptables -h' or 'iptables --help' for more information.

I read that this CONFIG_IP_NF_MATCH_OWNER has to be on in the kernel. Is it on for KP48?


EDIT: I saw that cmd-owner is removed from current version. Is there an alternative to it?