![]() |
Re: [Announce] kernel-power stable v52 in Extras
Here's my SR-enabled profile, which gets me noticeable lags during video playback. As an alternative I currently use non-overclocked non-modified ulv profile, lags very rarely. Same video, of course.
Code:
# SR-enabled non-overclocked |
Re: [Announce] kernel-power stable v52 in Extras
I must say I find it strange that you get lags with 720p libs installed (another post - are they still installed?), kernel power and if you use the dsp profile provided (sr enabled, slight overclock), yet not so much with a non-overclocked ulv profile.
Does it lag right after a reboot? This is likely o/t for kp. |
Re: [Announce] kernel-power stable v52 in Extras
Do you have swap on SD? That might help, also can you post output of iotop(in the repos) and top, is there anything to be found?
|
Re: [Announce] kernel-power stable v52 in Extras
Quote:
Quote:
I don't see a reason to use dsp profile, because ulv's performance is fine. I currently don't overclock my device. I will look into it later, once I've finally decided whether there is a way to make SR work properly on my device or not. Again, I have no performance problems when SR is off. I do have performance problems when SR is on. No matter rebooted or not, when SR is on I have noticeable lags during video playback. As far as I understand, SR main feature is to provide better battery life without decreasing performance. Well, that doesn't work for me. Also, I have Swappolube installed with proposed options stored, don't know if it matters. |
Re: [Announce] kernel-power stable v52 in Extras
SR does the undervolting and extra undervolting can cause performance hits. Did you use SR and undervolting together?
|
Re: [Announce] kernel-power stable v52 in Extras
Quote:
My SR profile is here, there is no undervolting there - it is just plain default kp profile with SR enabled. As far as I noticed, xlv profile undervolts lower than SR working on my device, but it doesn't bring any performance hits. |
Re: [Announce] kernel-power stable v52 in Extras
Just reinstall kernel-power-settings to get the all the default profiles including the one named 'default'.
Code:
sudo gainroot |
Re: [Announce] kernel-power stable v52 in Extras
You are talking about 'default' profile and that you messed it up.
Just to be sure (as it was not mentioned): You are aware that there are two occurences of default (and all other profiles)? One at /usr/share/kernel-power-settings/ and once more at /home/user/.kernel/ Possibly you messed up the one in /home by saving but 'restored' the one in /usr? The one in /home have higher priority (are in front of the search path). Please check also the contents of /etc/default/kernel-power (the profile loaded at boot). |
Re: [Announce] kernel-power stable v52 in Extras
First of all, thanks everybody for trying to help.
Now, back to the topic. I don't have a profile named default in /home/user/.kernel/. I do have a profile named default in /usr/share/kernel-power-settings/ and it is not messed. I thought I messed something because didn't realise at first that kernel-config was showing voltages with SR applied AND that SR is enabled in default profile... So when I saw undervolted and SR-enabled output of kernel-config right after loading default profile I thought omg wtf... However, I didn't really mess anything. /etc/default/kernel-power matches ulv profile, because I set it so. SR AND undervolting were used together in custom profile I deleted, just in case. I've tried again and again and I still get lags when using default profile and no lags using ulv profile, both non-overclocked non-modified standard profiles from kernel-power-settings. I'm out of ideas, especially regarding the fact I seem to be the first with this problem, and kinda not at all an experienced maemo user. |
Re: [Announce] kernel-power stable v52 in Extras
I guess that FlashInTheNight86 has a h/w peculiarity here. It is well known that various editions of the n900 behaved differently to SR enabled before FMG produced the current mostly stable implementation. However, his device likely is faulty but happily ulv seems to work.
As I too lack /home/user/.kernel I assume it is made for user modded profiles and not at installation. |
Re: [Announce] kernel-power stable v52 in Extras
Quote:
Code:
kernel-config save profile MyselfI don't even use kernel-setting at all and directly use echo commands to change parameters for on the go changing. Also I have edited pmconfig for a bootpersistant modification. Saves having this package installed. |
Re: [Announce] kernel-power stable v52 in Extras
My device is made in Hong Kong, so the problem indeed may be hardware-related...
Another question then - when overclocked to 900MHz on ulv profile, a couple times mplayer closed itself. Not a single stability problem otherwise. When overclocked to 900MHz on lv profile, there was no such thing. Do I get it right that this might be overclocked DSP-related issue, and I should add more voltage or reduce DSP frequency of some entries in ulv profile, or it still could be main CPU stability issue? |
Re: [Announce] kernel-power stable v52 in Extras
This now is firmly o/t. AFAIK, all production Nokia n900s were made in Korea (with a select number from Finland?), so yours is a "fake" (highly unlikely as it runs maemo) or a heavy rebuild. I would search the forum for user experiences...
|
Re: [Announce] kernel-power stable v52 in Extras
handaxe, I wrote made in italic to show my irony. It's a chinese refurbished N900 with new case and Made in Finland sticker, one of those sold on ebay. However, after reflashing it my experience is all positive, except some weird things (like not charging from some adapters and this SR problem). Anyway, good enough for 100$ I paid.
|
Re: [Announce] kernel-power stable v52 in Extras
Too subtle : )
|
Re: [Announce] kernel-power stable v52 in Extras
I remember that, when recalculating Sr, freemagordon was very thankful for some kind of output from all people, that experienced problems with recalculated values. Now, as you're first person in ages that have problems with SR (and seems to have everything set up *correctly*), I'm quite sure he would like to do d debug session with you. Unfortunately, he seems to be MIA for a few days, so I guess all we can do now is wait.
I'm, personally, very much interested in what causes your problems, especially that you have issues with SR recalculated values, that end up being *higher* voltage than ulv settings, where you *don't* have most of the issues. Strange. /Estel |
Re: [Announce] kernel-power stable v52 in Extras
Hello, does somebody have some pending patches for kernel-power? I still planning to release v53...
|
Re: [Announce] kernel-power stable v52 in Extras
what is the changelog of 53 as per u've did?
|
Re: [Announce] kernel-power stable v52 in Extras
current changelog for v53 is:
Code:
* Backport security fixes from upstream kernel for radio driver in patch radio-bcm2048-security.patch |
Re: [Announce] kernel-power stable v52 in Extras
Quote:
|
Re: [Announce] kernel-power stable v52 in Extras
noob - anything about battery life?
|
Re: [Announce] kernel-power stable v52 in Extras
EDIT: kp53 was released: http://talk.maemo.org/showthread.php?t=94287
|
Re: [Announce] kernel-power stable v52 in Extras
Hi Pali sorry to ask,but it's there pakect-injection for these modules?
regards |
Re: [Announce] kernel-power stable v52 in Extras
Quote:
..if yes, one thumb up for that! (my device was impatiently waiting for it) Eventually, it would be nice to add a new parameter in kernel-setting profile to load a custom voltage tweak directly from there if some device still needs it. (/sys/power/sr1_vdd1_dsp_boost) @Pali how about ext4 stability? Is there something that could still be patched to make it safer? |
Re: [Announce] kernel-power stable v52 in Extras
Quote:
|
Re: [Announce] kernel-power stable v52 in Extras
@pali
another still open problem is the usb host mode INstability (musb reset error etc..) But i have no idea how and if could be corrected. This is just a note in case you can take a look at this. Thanks |
Re: [Announce] kernel-power stable v52 in Extras
I have no idea how musb kernel driver can be improved/fixed...
|
Re: [Announce] kernel-power stable v52 in Extras
Quote:
|
Re: [Announce] kernel-power stable v52 in Extras
Quote:
/Estel |
Re: [Announce] kernel-power stable v52 in Extras
@Estel: I remember that you were able to reproduce problems with ad-hoc mode and packet injection drivers. Can you test new kp53?
|
Re: [Announce] kernel-power stable v52 in Extras
Quote:
|
Re: [Announce] kernel-power stable v52 in Extras
I actually started testing this while on train and hoped to report ASAP, but somehow, I managed to screw installation and I'm stuck on u-boot console with:
Code:
Booting attached kernel image... Code:
kernel-power-bootimg_2.6.28-10power53_armel.deb /Estel |
Re: [Announce] kernel-power stable v52 in Extras
This looks like damaged uboot mtd as uboot is not able to boot attached kernel. BTW, if you did not create custom combined uboot image, than attached image is not kp53...
|
Re: [Announce] kernel-power stable v52 in Extras
For the record - I've answered in u-boot thread:
http://talk.maemo.org/showpost.php?p...&postcount=756 ...as it's OT here and I don't want to clutter kernel-power thread. --- As for new wifi modules - I can (happily) report that ad-hoc wifi seems to be working perfectly fine now :) Apart from obvious lack of device reboot on trying to establish ad-hoc network, I was able to set-up one, connect to it from laptop, and share N900's cellular internet access through it. Same thing worked in reverse. Here good news end, though. As for power consumption, I did some tests - with everything disabled, screen locked and wifi ON but *not* connected to anything, power draw is the same for vanilla and injection modules. But, with device connected to managed network (access point) but idle (no VoIP accounts, no ssh, no anything) with vanilla drivers, device use ~15-20 mA. The same situation for injection modules results in ~200 mA current draw :( During heavy network usage and/or ad-hoc, vanilla and injection modules result in roughly the same power draw, so I suppose that some powersaving during idle is still lacking in replacement modules. This gives biiig regression re standby time, so using them instead of vanilla ones all the time doesn't seem to be such a good idea YET, after all :( /Estel |
Re: [Announce] kernel-power stable v52 in Extras
Ok, thank you for testing ad-hoc networks, its good to know that ad-hoc mode is working now.
When I'm using packet injection modules I see error messages about power save mode in dmesg. So this looks like problem... You can try to enable/disable power save mode via: $ sudo iwconfig wlan0 power on (or off) Anyway we still need to test if packet injection function still working. Patches which fixing ad-hoc mode probably disabling packet injection support when adapter is in ad-hoc mode. But this needs to be confirmed first... |
Re: [Announce] kernel-power stable v52 in Extras
Just a reminder to anyone volounteering to test it - don't trust aircrack-ng's injection test, I vaguely remember it lying that injection is working even, when it wasn't. Test it on actual (your own) WEP access point.
I would do the test myself, but I've, actually, purged everything WEP-related from my open-wrt powered router, and no WEP network in scanning range anymore, here (not that I would test it on someone's other network without their permission, of course ;) ). So, i must skip this test in someone's else favor. /Estel // Edit BTW, any changelog for kernel-power-settings18? |
Re: [Announce] kernel-power stable v52 in Extras
Quote:
Code:
* start upstart script after mounting /home and /opt |
Re: [Announce] kernel-power stable v52 in Extras
where I can find the changelog for kp53?
|
Re: [Announce] kernel-power stable v52 in Extras
Quote:
|
Re: [Announce] kernel-power stable v52 in Extras
Quote:
http://www.google.com/search?q=kerne...+v53+changelog Keep trying ... :cool: |
All times are GMT. The time now is 12:12. |
vBulletin® Version 3.8.8