Jump to content

An experimental CM7 Kernel


Guest t0mm13b

Recommended Posts

Guest t0mm13b
Hello,

I have recompiled the current CM7 kernel from nightly 14, what's the experiment you saw in the topic,

Well... I have enabled the options

CONFIG_ZTE_SUSPEND_WAKEUP_MONITOR=y

CONFIG_SCREEN_ON_WITHOUT_KEYOCDE=y
and fixed up the mis-spelling as per css171's post in another thread in relation to rpcrouter... ;) The above switches are enabled in the existing froyo releases, and seemingly battery drain is better on the froyo than the CM7, despite sharing the same kernel version 2.6.32... I am also hoping it fixes up wakelocks, AFAIK and from previous discussions with Tom_G, that its specific to the ZTE Blade... so.... Please do try it out and report here, whatever you do, PLEASE DO NOT bug the CM7 crew, nor HDCRJacob nor Tom_G on this... I am responsible so bug me instead... and refrain from posting in the other thread in respect to the RC2 nightly otherwise Victor will not be pleased to see nonsensical postings unrelated to RC2 :D The attachment when decompressed is boot.img and its md5 checksum is
d3a27ce2285fb7e8eebe50ac574841c0  boot.img

Edit:

BEWARE: that's a bad one.. wait... its missing modules... B)

Have phun and do report!!! \o/

Okey dokey... now sleepies time for moi! :D

Edit #2: (Heads up!)

Thanks to wbaw.. make sure you do the following: tick "Always use proximity" in call settings.

Update. 10th March, 2011, 21:00

Ok,

Heads up, this is a refresh, all debugging/unnecessary logging is stripped out of it... (heh! all your logging base belongs to us :D)

so I would be inclined to think it should speed it up slightly... ;)

Once we get satisfaction from users on their experience with this update then we'll use this as the base configuration to fix up other issues, and perhaps integrate andorko's FM radio ..... :D

Now, a simple matter of flashing t0mm13b_update2.zip via clockwork and all should be good to go... :o

Cheers. :D

Thanks to fonix232 for his guidance in helping me create the flashable clockwork script and to the gang in #zteblade! You know who you are :D

Update. 10th March, 2011, 23:00

Weird! :( have not changed the CONFIG_LOCALVERSION for the kernel and for some reason it will not load the wifi driver hence failure... fonix232 alerted me to it... I will look into the configuration again - perhaps I put the kernel on a very intensive dieting program to shrink it down from 3.2Mb to 2.7Mb :o.... Have withdrawn the zip file for the moment, investigating and recompile is on the agenda! ;)

Update. 12th March, 2011, 14:21

What a compile/test cycle I have been running :( there is definitely something weird going on with that darn wifi module ar6000.ko, which has exhibited a dependency on some kernel feature that I have switched off, have been recompiling at least 10 times tweaking the configuration, build the kernel, generate the clockwork script, reboot handset into clockwork, install, test... wash, rinse, repeat effectively, the wifi module requires the debugging features for some obscure reason which resulted in a breakage of the module load. :) Pity, that the ar6000 module cannot be recompiled, the date/time stamp on the wifi module is 1st August 2008... so that's saying something there...

I am still compiling quite literally as of now and will post out an update now that I have diff'd and merge'd the RC2's kernel configuration back into the tweaked kernel config - fingers crossed.

If this works, then rejoice! I will attempt to recompile the ar6000 module (the sources are something I would not be confident about) have tried to recompile it only for the compiler to complain and end the build of the module prematurely...

Will keep you all posted! :P

Update. 12th March, 2011, 16:00

w00t! working... here's the attached clockwork flashable update...

The source code for proximity/sensor data check can be found here via google code. For benefit of others, I have included the apk zip also so that you can check. Place the thumb near the sensor, it should say 'NEAR', take away the thumb, it should say 'FAR'. Then all should be go :D

Heads up!

Make sure you do the following: tick "Always use proximity" in call settings.

Experimental kernel update 21st March 2011:

Features:

  • FM Radio with Andorko's RDS patch supplied ... (BIG BUT!!! CM7 sources have dropped the fmradio support for some reason... it's there in codeaurora's sources - froyo to be exact) will be experimenting a few things there myself to see if I can rip out the sources from there and rebuild the android base framework code....
  • Wifi integrated into the kernel, no module...
  • Kallte Kaffe's patch for the headset
  • USB OTG... (anyone has the actual adapter for USB otg....?) this part is untested....

Due to the size of the kernel, the zip file is located on my dropbox a/c file: t0mm13b_exp.zip a/c.

Sensor_Application.zip

Edited by t0mm13b
Link to comment
Share on other sites

nice work ... except nightly 15 is the current version & maybe 16 will be released tonight, not sure if there were any changes to the kernel (seems unlikely, but i didnt check).

is the ahb overclocking stuff from jj9 in the cm7 blade kernel? i know it's not going to help battery life, but still, it'd be nice.

Link to comment
Share on other sites

Guest t0mm13b
So it doesn't work? we shouldn't install it? :(

Am a bit stuck here... :) in fact I am trying to create a simple update script to let clockwork install the boot.img and the kernel modules..... :(

Link to comment
Share on other sites

Am a bit stuck here... :( in fact I am trying to create a simple update script to let clockwork install the boot.img and the kernel modules..... :(

send me an email (check your pms) & i'll sort it out for you, i don't need sleep.

Link to comment
Share on other sites

Done the clockworkmod3 update for this, tested & working (it copies the files & the phone boots ok) - not sure if it'll help battery life or anything else, it needs testers!

It adds a lot of extra modules to the /system partition, so you'll need at least 4.5mb free on there to install it.

http://www.mediafire.com/file/311zq39vb9c2...m13b_update.zip

md5sum:

f1fad9d7fa7b44d60717e03c0e2ede29 t0mm13b_update.zip

Edited by wbaw
Link to comment
Share on other sites

Guest burstlam

thanks.

I would like to know more about these modules. can u explain more about their function and necessity?

I am now using RC2 with boot.img fix inspired by css171, after reading the threads concerning power drain issue.

I have switch off the data connection, wifi ,gps,auto sync. Removed rom manager. set wifi sleep policy to never. set end call action to home and sleep.

also with dewonkificator installed to deal with the blackout issue.

now power drain issue in sleep mode seems fixed to me. it drops less than 1% an hour in sleep mode, just like the 2.2.1 do.

however, power consume more in 2.3.3 compare to 2.2.1 when making calls or net surfing. my friend running a cm7 on nexus also encouter power drain problem... is it something relate to 2.3.3?

anyway, I will full charged the phone and test with ur mod.

again , thank you so much

Link to comment
Share on other sites

kernel modules are parts of the kernel which have been separated out into modules, they're often drivers. with this update the cifs module, which provides compatibility with windows file shares is 3mb rather than under 300k as standard in cm7, no idea why, but that's what takes a lot of the extra space. there are also some other extra modules, not sure what they do, a few seem to be different test modules. you'll have to wait until t0mm13b wakes up to explain exactly what they all do.

Edited by wbaw
Link to comment
Share on other sites

Guest Jekle

Instead of Fixing up the .32 kernel why not try and port the .37 or .35 kernel? If we port the CM7 .37 kernel it'll already has tweaks and optimizations for CM7 which may include battery life enhancements.

Link to comment
Share on other sites

it's too early to tell, i'm still at 92% battery, but it's lost a good few percent in the last 90 minutes of not doing much. so i'm not sure it has any effect on battery life, but the screen works as it should during calls!!! :(

Link to comment
Share on other sites

Instead of Fixing up the .32 kernel why not try and port the .37 or .35 kernel? If we port the CM7 .37 kernel it'll already has tweaks and optimizations for CM7 which may include battery life enhancements.

because this is easier. i agree it'd be better to have a .37 kernel, but it'd also be quite a lot more work. if you're volunteering, then go for it :(

Link to comment
Share on other sites

Guest norwici
i tested this too, and it doesen't work for me :( screen stays off like before.

The proximity behaviour seems much improved to me on an old OLED with RC2.

I will see how it performs today but it certainly looks like we might have grounds for some optimism.

Thanks for this.

Link to comment
Share on other sites

Guest hecatae
Instead of Fixing up the .32 kernel why not try and port the .37 or .35 kernel? If we port the CM7 .37 kernel it'll already has tweaks and optimizations for CM7 which may include battery life enhancements.

what effect would switching from a 2G/2G VMSPLIT to a 3G/1G VMSPLIT have on battery life?

Link to comment
Share on other sites

The proximity behaviour seems much improved to me on an old OLED with RC2.

I will see how it performs today but it certainly looks like we might have grounds for some optimism.

Thanks for this.

works for me on a uk tft christmas 2010 vintage, on nightly 13, 100%, if i have 'Always use proximity' enabled in call settings. without that enabled, it's not so good.

Link to comment
Share on other sites

Guest t0mm13b

Ok! and Good morning...

kernel modules are parts of the kernel which have been separated out into modules, they're often drivers. with this update the cifs module, which provides compatibility with windows file shares is 3mb rather than under 300k as standard in cm7, no idea why, but that's what takes a lot of the extra space. there are also some other extra modules, not sure what they do, a few seem to be different test modules. you'll have to wait until t0mm13b wakes up to explain exactly what they all do.

Kernel modules are really what would be from a Windows world - drivers, nothing more really, they provide the kernel the means to interact with whatever hardware is in place and respond and request data to it. As for the cifs module - I know why it's huge - there's unnecessary attributes compiled in which is defined as default within the CM7 kernel... don't ask me why but on the face of it, its not needed for simple file sharing....

it fixes the proximity sensor bug, for me at least :)

WHA' ? :(

Nice stuff t0mmi3b and wbaw. Gonna try this too tonight.

it's too early to tell, i'm still at 92% battery, but it's lost a good few percent in the last 90 minutes of not doing much. so i'm not sure it has any effect on battery life, but the screen works as it should during calls!!! ;)

No way josé, that's friggin impossible!!!! :(

Sheesh, after reading this made coffee come out of me nostrils and sprayed the screen in a orderly fashion... meh.... that's some surprising results from wbaw... ;)

A big kudos and thanks to wbaw for helping me out with the update script! Cheers!

Please do report back and see... have a funny feeling that the CONFIG_ZTE_SUSPEND_WAKEUP_MONITOR activates other parts of the code elsewhere and not just for monitoring wakelocks, perhaps ZTE specific stuff... no idea... not 100% sure as I only gleaned in on the code... will investigate this further...

Edited by t0mm13b
Link to comment
Share on other sites

Guest Rotmann
So far there's one person saying it didn't fix the blank screen... But if it even lowers the chance of a freeze, it's good :(

That person does not say if "Always use proximity" was ticked.

The first post should be updated with wbaw's version and mentioned that people should tick "Always use proximity".

Link to comment
Share on other sites

What exactly does this "always use proximity"-option do?

it says 'enable to let the proximity sensor lock the screen even if phone is horizontal'. in practice, it makes it work. i had that enabled before, unticked it tested it again & it was broken like before the new kernel, ticked it & it works again. i think it might be off by default in cm7.

Edited by wbaw
Link to comment
Share on other sites

battery use still seems high, i'm down to 78%, but i don't let my phone totally idle much anyway. that will need testing by somebody who lets it idle, mine is listening for voip calls, checking emails, keeping me logged into chat, checking the weather & the news, wifi on, etc, etc, etc.

next job, t0m, fm radio please :(

Link to comment
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.