[KERNEL] [Y300/G510] Stock Huawei

[Guest]

Ok, that's weird. I didn't expect any screen related issues with this build at all.

Can you confirm the build date of Jul 30 21:36:24 CEST 2014 if you check /proc/version?

It seems we got different hardware, although we got the same device :-/

Going to recompile the kernel this afternoon.

Can anyone else confirm this issue, too?

[Guest]

@SH3H1: a 3.4.y kernel will never become 3.5.y after patch release, because it already exists :P

Despite the version of the synopsis kernel is getting shown as 3.4.0, it got a couple of updates uncluded, taken and ported from higher versions: 3.5-3.17

[Guest]

i use trickster mod

but 4 months ago :D

The problem is, that only "voltage levels" are getting exported and not the real amount of mV being used. This doesn't allow any fined grauned control of the voltage itself.

[Guest]

Can someone please check if the freezing issue is gone now: http://www22.zippyshare.com/v/2124625/file.html

Edited July 31, 2014 by Guest

[Guest Domino2115]

sure, wait

edit: it still dont get up from deep sleep :/

Edited July 31, 2014 by Domino2115

[Guest]

Hm ... so it's not display driver related

[Guest fonz93]

i tried both versions and i have the deep sleep issue too

[Guest SaurabhRCk]

I tried the latest one! I dont have any display issue on lock screen. but my device reboots automatically after 10 mins of standby. I reverted to old 25th build! Now its ok!  

[Guest]

On which ROM(s) did those issues occur?

[Guest Domino2115]

On which ROM(s) did those issues occur?

B209 cleaned by fonz

[Guest fonz93]

On which ROM(s) did those issues occur?

 

I am on H3ROS ROM (AOSP MOD)

[Guest SaurabhRCk]

I am on gamer r1 by jaccs. (b200)

[Guest squal]

me too..im in r1_gamer b200 by jaccs, deepsleep issue too, but i reverted to synopsis 3.4.0 july 24 build works fine, i can play dead trigger2 ( medium performace) subway surfer and virtua tennis challege, no lags and no deepsleep issue:)

Edited July 31, 2014 by squal

[Guest]

Hmm. I've installed an stock B199, untouched, and only removed some system apps and made some customizations.

[Guest]

experimental test release: http://www49.zippyshare.com/v/31348085/file.html (GCC 4.9.1)

 

[ GCC 4.8.1 build: http://www72.zippyshare.com/v/16786884/file.html ]

 

APP for tracking CPU-usage in statusbar: https://play.google.com/store/apps/details?id=jp.takke.cpustats

 

Changes:

• CPU hotplug is getting controlled by the kernel itself (enabled by default)
• removed MPDecision service (so it doesn't/can't get started on accident)

Known issues:

• accessing the flashlight is a bit delayed. still need to find the reason and fix it.

ToDo

• fine tune the new kernel's hotplug driver

 

A small feedback about the hotplug behavior would be nice. Also let me know if the screen still freezes.

 

 

NOTE: this kernel build is not supposed for all-day-usage yet!

 

Tuneable I am able to adjust are:

    .delay = 100,
    .scroff_single_core = true,
    .max_cpus = 2,
    .min_cpus = 1,
    .cpufreq_up = 1000000, //kHz
    .cpufreq_down = 700800, //kHz
    .cycle_up = 1,
    .cycle_down = 3,

Edited August 1, 2014 by Guest

[Guest Domino2115]

15minutes in deepsleep, no problem with waking up, good work

[Guest pbr1111]

This commit in the Dazzozo's github could be an improvement?

https://github.com/Dazzozo/android_device_huawei_msm7x27a-common/commit/f218c7914fa8752026b021cf04cff7b7a03aad5f

[Guest]

When compiling a ROM it could. Kernel's don't benefit from it, because all kernel modules are using INTEGER math. No floating-point operations. No chance to make use of the FPU (floating point unit). Thx for your suggestion anyway ;)

[Guest SH3H1]

My phone is too sleepy that it even forgets to turn off the led light while charging sometimes (screen off ~10 Min) :P

[Guest]

this build is targeted towards experienced users only (for testing the new auto-hotplug feature)

 

updated experimental test build: http://www18.zippyshare.com/v/6572418/file.html

 

• kernel got auto-hotplug (doesn't make use of mpdecision anymore)
• small updates for ZRAM
• squashed some block driver updates from 3.4.101

[Guest]

Of course that build got tested and is fully functional :)

It just needs some fine-tuning, because the auto-hotplug feature is not behaving perfectly as imagined, yet. But it's still better than MPDecision and more energy-efficient.

[Guest Romagnolo1973]

Probably are all in holiday here in Italian Riviera, sun beach and disco :-)

 

Testing OnDemand + Sio: 

- boot perfectly

- it works 100% smooth

 

So now the MP decision part in Performance Control must be unticked? I have deselect it

How can we help? Using CPUSpy and check cpu usage?

[Guest]

Since I removed the definition for the MP-Decision service from the scripts of the ramdisk, it shouldn't be possible that it gets started on accident, unless something else starts it on accident.

To be on the safe side it's not a bad idea to rename the mpdecision binary. But it should not be necessary.

Things required to observe now are:

- battery drain

- performance on demanding workloads (ex. gaming)

- CPU hot-plug behavior in general

[Guest]

My average battery drain is around 2-3% per hour. This includes phoning, sending/reading text messages and browsing the web from time to time. Sometimes I listen to music for a few hours per day, too, which increases my average into the 3-4%/hour range.

With the new auto-plug I personally experience an slight improvement.

But this obversation needs to be treated with caution, since most of my average workloads are not very demanding.

Would be nice to get some feedback of you testing stuff like gaming, video playback, etc.

[Guest Domino2115]

Battery is great i really like it, from 99%-92% in 2 hours and i was browsing the web, fb messenger, tapatalk, playing subway surfers and listen some music. I tried subway surfers with older builds it laggs from time to time but now its totaly smooth, also overall performane/smoothnes increased for me.

Edited August 6, 2014 by Domino2115