[KERNEL][GPL][N4] franco.Kernel - r217

Search This thread

pa.pn2

Senior Member
A

asasione

Guest
Did that aswell. App works but just some settings are missing.
Never had such before
Problems started with fku update to 10.0.5

That's francisco's department now, no idea, when I said full wipe, hope you formatted data (factory reset) and system and started clean
 

fkloznica

Senior Member
Apr 16, 2013
712
278
Noob q: can i activate all 4 cores on this kernel ? On trickster mod and cpu z i see only 2 cores working
 

Arstoien

Senior Member
Oct 31, 2012
1,438
1,508
Noob q: can i activate all 4 cores on this kernel ? On trickster mod and cpu z i see only 2 cores working

And you can control how "easy" they are activated by changing the high load threshold in Franco's app, or assuming you don't have it, change the value of this file (it's a number from 1 to 100, showing the load that's needed for all cores to online)

/sys/class/misc/mako_hotplug_control/first_level

Sent from my Nexus 4
 

aventu

Senior Member
Apr 17, 2011
413
786
@franciscofranco

I get 3 warnings when I try to compile the source from github

Code:
WARNING: vmlinux.o(.text+0xab20): Section mismatch in reference from the function mako_hotplug_late_resume() to the function .cpuinit.text:cpu_up()
The function mako_hotplug_late_resume() references
the function __cpuinit cpu_up().
This is often because mako_hotplug_late_resume lacks a __cpuinit 
annotation or the annotation of cpu_up is wrong.

WARNING: vmlinux.o(.text+0xad2c): Section mismatch in reference from the function decide_hotplug_func() to the function .cpuinit.text:cpu_up()
The function decide_hotplug_func() references
the function __cpuinit cpu_up().
This is often because decide_hotplug_func lacks a __cpuinit 
annotation or the annotation of cpu_up is wrong.

WARNING: vmlinux.o(.text+0xad9c): Section mismatch in reference from the function decide_hotplug_func() to the function .cpuinit.text:cpu_up()
The function decide_hotplug_func() references
the function __cpuinit cpu_up().
This is often because decide_hotplug_func lacks a __cpuinit 
annotation or the annotation of cpu_up is wrong.

The kernel won't boot if I ignore the warnings. My setup is new, so it could be my fault.

edit: I've done something wrong, sorry for that :/
edit2: It still doesn't work, I fixed a bug with packing the boot.img, but the kernel doesn't boot
 
Last edited:

Cantacid

Senior Member
Oct 4, 2012
205
52
Stop the damn music, change the value wait a few seconds and play again. Thats what says in the option on the app.

+1 for the mic! My gf has been complaining about this too lol when they're pissed, they're pissed indeed :p

Sent from my Nexus 4 using xda app-developers app

Haha my gf has been bugging me about since the day i got this phone. Thought i did not have a way out. Franco, go for it.
 

thederekjay

Inactive Recognized Developer
Nov 15, 2010
4,628
8,913
DARKSIDE
@franciscofranco

I get 3 warnings when I try to compile the source from github

Code:
WARNING: vmlinux.o(.text+0xab20): Section mismatch in reference from the function mako_hotplug_late_resume() to the function .cpuinit.text:cpu_up()
The function mako_hotplug_late_resume() references
the function __cpuinit cpu_up().
This is often because mako_hotplug_late_resume lacks a __cpuinit 
annotation or the annotation of cpu_up is wrong.

WARNING: vmlinux.o(.text+0xad2c): Section mismatch in reference from the function decide_hotplug_func() to the function .cpuinit.text:cpu_up()
The function decide_hotplug_func() references
the function __cpuinit cpu_up().
This is often because decide_hotplug_func lacks a __cpuinit 
annotation or the annotation of cpu_up is wrong.

WARNING: vmlinux.o(.text+0xad9c): Section mismatch in reference from the function decide_hotplug_func() to the function .cpuinit.text:cpu_up()
The function decide_hotplug_func() references
the function __cpuinit cpu_up().
This is often because decide_hotplug_func lacks a __cpuinit 
annotation or the annotation of cpu_up is wrong.

The kernel won't boot if I ignore the warnings. My setup is new, so it could be my fault.

Toolchain. Use 4.6. Or fix the warnings.
 

Net.silb

Senior Member
Nov 24, 2011
1,042
771
Noob q: can i activate all 4 cores on this kernel ? On trickster mod and cpu z i see only 2 cores working

And you can control how "easy" they are activated by changing the high load threshold in Franco's app, or assuming you don't have it, change the value of this file (it's a number from 1 to 100, showing the load that's needed for all cores to online)

/sys/class/misc/mako_hotplug_control/first_level

Sent from my Nexus 4

Afaik it's from 0 to 100 not 1 to 100, so if you want all 4 cores on all the time (screen on of course...) just set high load threshold to 0 and you're good to go. If you want only 2 cores on all the time just set it to 100. Otherwise, as @Arstoien said the higher that value is, the "harder" it will be for 4 cores to be online.
 
Last edited:
  • Like
Reactions: fkloznica

thederekjay

Inactive Recognized Developer
Nov 15, 2010
4,628
8,913
DARKSIDE
I use arm-eabi-4.6 (Google 4.6 Toolchain) like Franco.

edit: I've done something wrong, sorry for that :/
edit2: It still doesn't work, I fixed a bug with packing the boot.img, but the kernel doesn't boot

Hmmmm.. What distro are you using? Do you have your packages for your build environment setup correctly??
 
Last edited:

Top Liked Posts

  • There are no posts matching your filters.
  • 2376
    Hi all.

    Current builds (starting with r217) are for Lollipop 5.1 only.
    Builds (starting with r214 up to r216) are Android 5.0 Lollipop only.
    KitKat is dead and the last stable release was r213, which you're free to use if you're still on 4.4.4.

    Download:
    boot image:
    http://kernels.franco-lnx.net/Nexus4/5.1/
    zips:
    http://kernels.franco-lnx.net/Nexus4/5.1/zips

    Installation:
    fastboot flash boot boot_name_img
    or
    flash the .zip in the recovery

    Source:
    https://github.com/franciscofranco/mako/tree/lollipop-mr1

    Changelog:
    http://kernels.franco-lnx.net/Nexus4/5.1/appfiles/changelog.xml

    franco.Kernel updater Free apk: http://xdaforums.com/showthread.php?t=1867127

    Got any questions that you couldn't find after searching?
    FAQ: http://xdaforums.com/showthread.php?t=2532422
    197
    Color Calibration for the Nexus 4

    Thanks for all those of you who provided feedback on my color calibration settings using r72 and Franco's app.

    Someone messaged me to ask what the thinking was behind my settings, and I thought it might be worth posting the reasons here too so everyone can understand the why behind them.

    Because this thread moves so quickly, here are the settings I'm using again, so you don't have to go searching for them:

    R 175
    G 175
    B 175

    Whites 20
    Mids 100
    Blacks 255
    Contrast 0
    Brightness 0
    Saturation 200
    Grays 0

    Screen Brightness 50%

    These settings provide a VERY close approximation to my calibrated PC monitor when studying my photos in detail (I'm a cinematographer and photographer by profession - see here: http://www.visceralpsyche.com).

    A lot of you love the settings, but some of you also say they are too dark, or wonder why you can't put a lower screen brightness and compensate by increasing the RGB values to let more light through.

    To understand why I chose my settings, you have to remember that our displays are made up of two parts - the backlight (screen brightness control in Android) and the LCD panel itself (adjusted by Franco's app), which the backlight shines through.

    What I found with my testing was that with RGB set to 255, specular highlights (and white edges in general) tended to blow out, causing color fringing artifacts. This is easily seen as pink or yellow edges to what should be pure white images. As I pulled back the RGB values, this fringing began to get smaller, until by my settings of 175 there was no longer any color fringing to my whites. Think of it as a form of highlight recovery. By pulling the clipped channels down equally, I found the point at which the values didn't overflow the hardware limitation of the LCD panel itself.

    Once this was done, I went into the gamma interface and began looking for where the whites and blacks ran close to the limit, but not over. I adjusted the screen brightness to where I could distinguish the bottom and top values as individual shades. THis was close enough to 50% that I set it to 50% and tweaked it to be perfect.

    Then I played with the Mid and Gray values until middle grey was indeed in the middle, so that the tone curve wasn't biased to one direction.

    From there, I adjusted saturation to match my PC monitor's saturation level.

    After all that, I arrived at the values I posted above, and now have a screen which faithfully replicates the tonality of my photos without any artifacts or strange tone curves.

    If you want to test these values (or your own) I recommend the following images. The calibration chart is one I found online a long time ago, which is useful for seeing the top and bottom end of white and black, to make sure your screen is showing you all the proper values:

    calibration-chart.gif


    The photo is my own, and is useful because I know exactly how I processed it and therefore that it touches both the white end of the spectrum (the highlighted side of her face, with values at precisely full white) and the black end of the spectrum (the fine detail in her hair, which is very black but not crushed in the details):

    IMG_1362a_framed_sRGB_25.jpg


    You can also use her facial tonality to make sure you see a smooth transition through the grays, not steps.

    It's also black and white, which means that any color fringing you see is entirely due to artifacts introduced by errors in the color or gamma settings and not the backlight of the screen).

    Both images are hosted on my own server, so there are no compression artifacts introduced by the forum.

    Again - my aim with these settings is to give you a proper, reference screen that displays accurate colors without any artifacts or errors. Brightness is set to match a proper reference display (120cd/m2) and may be brighter or darker than you prefer.

    If you value absolute battery savings over accurate color, then these settings may not be for you. That's perfectly ok of course. But for me, in my line of work, I value accuracy over the little bit of battery savings I might get by lowering my screen brightness, so if that's your goal too, then you'll find these colors are a good basis for having the best possible photo viewing screen on any phone.

    Because I can tell you, I have never had a more accurate screen than the Nexus 4 when it comes to looking at my photos, when combined with Franco's fantastic color interface and kernel.

    Hope this helps!

    Cheers,

    Paul Leeming
    Writer/Director/Cinematographer
    Visceral Psyche Films
    http://www.visceralpsyche.com
    162
    Really sad to read some posts after all the work I put on this device for longer than an year. Anyway here's r204 that I've been working on for a while...

    * Introduce a new bunch of tunables for my Hotplug driver and squash all code improvements from Hammerhead's version
    * Uses a power efficient system workqueue to allow the cores to stay longer times on idle
    * Several key subsystems are now using a newer and more advanced kernel thread infrastructure to allow cores to allow for better power savings
    * Improved thermal throttle code
    * Merged all upstream 3.4.y patches
    * Merged misc patches

    Have fun.
    157
    I couldn't resist. My girlfriend is a bit mad at me, but I told her I would buy her some nice drinks after dinner so while she's happy I managed to get the new Kernel sources, build fresh with the new ramdisk and I only had time to merge my custom changes:

    - Cleaned config from crap
    - Added my Hotplug driver (bye mpdecision)
    - Added my custom changes to the Thermal driver (bye thermald)
    - Added my custom changes to Interactive Governor (and made it default)

    And thats pretty much it. I don't have more time now, (we'll see tomorrow, I rented a hut in the beach for tomorrow for me to work a bit on the Kernel/App while getting some nice sun and my GF won't get mad eheh - Android, GF, drinks, sun and beach, who doesn't love this. I'll take a picture tomorrow and share with you all). Attached is r170 for 4.3 with the changes I mentioned above.

    Have fun guys.
    140
    Ok I have a small gift for all of you that run FK for almost a year on this device. I always said that I would never implement any of those functionalities to wake the device without pressing the power button - point is that LG G2 has brought this feature, Oppo N1 has something very similar and... I've been getting emails that users are having the Power Buttons wore off. Initially I implemented this on the new Nexus 7 as a challenge and to have some fun. Couple code lines later I had the functionality working fine and dandy. The device deep sleeps and the only thing that is being kept active is the digitizer for this to work. I sent it to a couple of people and they enjoyed it specially because it worked better than other Kernels that had implemented similar options. Good, so lets try and implement it on Mako. It's 100% my code, pretty simple, and easy to understand. Please refer to git (I just pushed the code) for how it works in the touch driver and input subsystem.

    By default this option is disabled, but you can enable it by doing:

    echo 1 > /sys/module/lge_touch_core/parameters/touch_to_wake

    After this is enabled you can power up the display by double tapping just like the LG G2, and you can swipe from whenever you want on the display and it will power up. There is a 100ms (0.1s) minimum interval where the algorithm will detect touch events so filter ghost touches and some unintended shít. It seems very low value, but when we're touching the screen the digitizer is capturing hundreds of touch events every minute, and dozens every second. So this works as a filter.

    If you're using touch_control and you want to test this, please disable the module and reboot your device or you'll most likely get into some **** problems. And the same thing other way around.

    Oh and interactive is back to default, it just gives me more consistence results all around.

    Attached goes two test zips, flash, enable the touch thingy, test, and report back.