Aller au contenu

[Kernel] [4.4.x/L] Hell's Core b56


Recommended Posts

Bonjour,

 

Je crée se topic, pour tenir informer nos chers Nuxiens sur l’évolution et les mises à jours du Kernel Hell's Core.

Je ne suis pas le concepteur du Kernel, j'informe juste et met à disposition les liens nécessaire.

 

233680d1377019913-kernel-jb-jss15j-jwr66

Pour l'avoir testé, il me tien facile 1 journée voir l'autre d’après en utilisation normal(avec un peu de net, dl...)

Le téléphone ne chauffe pas, et chargement rapide.

 

Téléchargement :

 

Hell's Core b56/b56 CM

Hell's Core b56/b56 CM (miroir)

 

Application pour gérer le Kernel --> Hell'score kernel manager 3.1

 

Change Logs:

 

(b56)

-merged some upstream fixes

-reverted two cpuidle patches

(b55)

(b54)

-New governor: hellsactive*

-speedup late resume (faster wakeup)

-ext4 patches

-pm/bus patches to prevent reported crashes (feedback needed!)

-some other little fixes here and there

-compiled with boostedtc 4.9.2 by [MENTION=2607869]aznrice2k4[/MENTION]

*hellsactive:

-Based upon intelliactive

-removed input boost (we have an input boost in the hotplug driver)

-removed sync feature

-removed sampling down factor

Following was added:

-Align timer windows for all CPUs*

It's more advantageous to evaluate all CPUs at same time so that interactive governor gets a complete picture of the load on each CPU at a specific time. It could also reduce number of speed changes made if there are many CPUs controlled by same policy. In addition, waking up all CPUs at same time would allow the cluster to go into a deeper sleep state when it's idle.

-Reset last evaluated jiffy in idle callback*

When CPU has been busy for a long time, last evaluated jiffy will be quite behind because the timer would have been canceled. We don't want to schedule a timer to fire in the past as load will always be 100%. Reset last evaluated jiffy so that timer will be scheduled for thenext window.

-Round up timer_rate to match jiffy*

Timers are scheduled in unit of jiffies. Round up timer_rate so that it matches the actual sampling period.

-Use del_timer/add_timer_on to rearm timers*

Replace mod_timer_pinned() with del_timer(), add_timer_on(). mod_timer_pinned() always adds timer onto current CPU. Interactive governor expects each CPU's timers to be running on the same CPU. If cpufreq_interactive_timer_resched() is called from another CPU, the timer will be armed on the wrong CPU.

Replacing mod_timer_pinned() with del_timer() and add_timer_on() guarantees timers are still run on the right CPU even if another CPU reschedules the timer. This would provide more flexibility for future changes.

-Support variable frequency calculations**

Use a threshold value up to which we are evaluating the current speed weighted by current load against each new choice of speed. Above switch to the original frequency calculation proportional to load relative to the max possible frequency which allows for more dynamic and fair changes.

Default threshold is policy->min which guarantees minimal freqs and energy usage during (offloaded) media playback.

Also, choose the highest frequency below or at target, which generally saves us one freq level.

-Don't scale beyond min if the load is low***

-Thread migration commit***

http://forum.xda-developers.com/showpost.php?p=55298804&postcount=7952

I've just made some little changes on commits to suit my needs.

credits:

*CAF

**myfluxi

***franciscofranco

(b53)

-fixed cores getting stuck at input_boost if you've min 2 cores online: commit

-updated msm_mpdecision with latest nexus 5 commits

(b52)

Compiled with Linaro 4.9.2-2014.08 by Christopher83

New build flags from my hammerhead branch and [MENTION=2923435]barome[/MENTION]

Code is identical with b51. Just a newer toolchain and new build optimizations. Let's see if there are new stability issues. I think we've gained some little performance boost

(b51)

-Added some commits for the camera. We had some reports about reboots while using flash. Maybe there'll be less of them. I don't think I can fix them completely...

-Removed some debugging

-Some love for conservative governor (thx stratosk)

-Added KERNEL_MODE_NEON

-Changed some little things in the flags

-Some little things here and there. Refer to git if you'd like to know it exactly

(b50-test)

-No problems occured, just merged some cool stuff

-Merged backlight dimmer from stratosk

-Added a ton of kgsl/adreno patches (L sources) and Franco`s interactive GPU governor

-Added UIO Sharedmem (L sources)

(b49-test)

NO UNDERVLOTING ANYMORE , stock voltages by default

Compiled with Linaro GCC-4.9.1 from Christopher83

msm_mpdecision from showp

intelliactive governor (default)

ZEN scheduler (default)

FIOPS scheduler

Many many things under the hood (check git)

Touch Wake (stratosk)

DT2W (stratosk)

and so on... ... ...

http://forum.xda-developers.com/showpost.php?p=55061297&postcount=7579

(b47-pre-release) STABLE

Merged 4.4.3

Update to Linux 3.4.93

Linaro 4.9.1 from Christopher83

IMPORTANT NOTE!!!!!

IF YOU COME FROM HELLSDOCTOR, PLEASE REFLASH YOUR ROM OR CLEAR YOUR INIT.D FOLDER!!! OTHERWAYS YOU'LL HAVE FC's LIKE HELL!

(b46 test 4)

-OOM Upstream to 3.14 from faux

-S2W Fixes from AK

-Linux 3.4.86

(b44-45)

-Upstream auf 3.4.84

-Linaro 4.8.3 03.2014

-Cpuidle Commit

-futex Upstream updated

-jbd2 commits optimized

-LoUIS Cache

-81mhz

(b44 test t3)

Lowest frequency 81 MHz

Updated to linux 3.4.79

Merged CM optimizations

(b43)

-Rebase auf 4.4.1 Sourcen

-Linux 3.4.76

-S2W/dt2w Port vom N5

-faux sound 3.2

-suspend_freq auf max angehoben

-Kleine Fixes usw. siehe Git

(b40)

-Upstream from 3.4.65 auf 3.4.66

-ca. 100+ SLAB/SLUB/SL[AU]B/mm... Commits

-Module Support (for TouchControl)

-Conservative GPU: Thresholds set slightly differently ( probably a little smoother when scrolling)

-S2W Fix from fluxi

-kexec Support (MultiROM)

-3 epoll Commits from Upstream

(b39)

-Rebase auf CM/JSS/JWR Sources

-Compiled with Linaro 4.8.2.09 Toolchain

-Upstream from 3.4.65

-O3 Support with Custom Build Flags

-Conservative GPU Governor

-msm_hotplug Hotplug Control (Tanks geht an fluxi)

-dynamic fsync off

Lien vers le commentaire
Partager sur d’autres sites

Archivé

Ce sujet est désormais archivé et ne peut plus recevoir de nouvelles réponses.

×
×
  • Créer...