[KERNEL][GPL]2013-03-02 b24 motley kernel for the Prime (4.2.2 alpha)

• Forked from namidairo (thanks!)
• OC to 1.7GHz
• GPU OC to 600MHz - configurable from 520-650MHz (see below)
• Voltage Control (use Trickster or System Tuner)
• Removed system_mode from Asus - boots to 1.4GHz - control with standard apps (use Trickster or System Tuner)
• FSYNC - 3 modes (see below) (use Trickster)
• Linaro built with optimizations
• Kernel log spam reduced - important for our crappy eMMC
• optimized ARM RWSEM algorithm
• Other misc tweaks (see github)

• Initial release for 4.2.2
• AnyKernel install works on 32GB Primes only
• Added blob based installs for CM 10.1 and PA. Read notes carefully below when choosing an install zip.

echo 520 > /sys/devices/system/cpu/cpu0/cpufreq/gpu_oc

echo 650 > /sys/devices/system/cpu/cpu0/cpufreq/gpu_oc

echo 1 > /sys/kernel/dyn_fsync/Dyn_fsync_active
echo 1 > /sys/class/misc/fsynccontrol/fsync_enabled

echo 0 > /sys/kernel/dyn_fsync/Dyn_fsync_active
echo 0 > /sys/class/misc/fsynccontrol/fsync_enabled

echo 0 > /sys/kernel/dyn_fsync/Dyn_fsync_active
echo 1 > /sys/class/misc/fsynccontrol/fsync_enabled

#!/system/bin/sh 
echo 4 > /sys/module/cpu_tegra/parameters/system_mode
echo 1350 > /sys/kernel/tegra_cap/core_cap_level
echo 0 > /sys/kernel/tegra_cap/core_cap_state
echo 0 > /sys/devices/platform/tegra_grhost/enable_3d_scaling
echo 10 > /sys/module/cpu_tegra3/parameters/mp_overhead
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_1
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_2
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_3
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_4
setprop power.saving.mode 1
setprop persist.tegra.NV_FPSLIMIT 0
setprop persist.sys.NV_FPSLIMIT 0
setprop persist.sys.NV_POWERMODE 0

echo 1700000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
echo N > /sys/module/cpu_tegra3/parameters/no_lp

#!/system/bin/sh
# auto zram activation init script with busybox search
# by show-p1984

echo "[90ZRAM]: Firing up /system/etc/init.d/90zram";

if [ ! -e /sys/block/zram0/disksize ] ; then
        echo "[90ZRAM]: ERROR unable to find /sys/block/zram0/disksize";
        echo "[90ZRAM]: Is this a ZRAM kernel?";
        echo "[90ZRAM]: ZRAM NOT ACTIVATED. (404)";
else
        #find busybox in /system
        bblocation=$(find /system/ -name 'busybox')
        if [ -n "$bblocation" ] && [ -e "$bblocation" ] ; then
                echo "[90ZRAM]: busybox found in:" $bblocation;
                echo "[90ZRAM]: Setting ZRAM disksize.";
                echo $((100*1024*1024)) > /sys/block/zram0/disksize

                echo "[90ZRAM]: Starting ZRAM...";
                bblocation=${bblocation%/*}
                cd $bblocation
                ./busybox mkswap /dev/block/zram0
                ./busybox swapon /dev/block/zram0

                echo "[90ZRAM]: ZRAM activated.";
        else
                echo "[90ZRAM]: ERROR! busybox not found!";
                echo "[90ZRAM]: Is busybox installed? Symlinks set?";
                echo "[90ZRAM]: ZRAM NOT ACTIVATED. (404)";
        fi
fi

• alpha version, experienced users and testers only please!
• Do a CWM backup so you can restore your boot.img and ROM if necessary
• 64GB tabs please use blob based install with ramdisk. Anykernel method may not work on 64GB tablets
• Recommend 1.6 or 1.7GHz. I have scored 14300+ on Antutu with 1.7GHz. 1.8 is just not necessary IMO and generates too much heat.

• Still using stock broadcom WiFi drivers (module built by me match stock modules)
• Still use same scaling scripts from before (cpu3.sh etc.) or installed 2.6.11 first and then flash this kernel to be sure you have them right.
• Same system modes apply as before (system_mode=3 for 1.6GHz, system_mode=4 for 1.7GHz, system_mode=5 for 1.8GHz etc.)

• Supports Gummy, Team EOS, and CM9 since they have the same WiFi driver implemented in their device trees.
• No longer need to use an OC app (SetCPU, No-frills etc.) to set the clock.
• WiFi driver is built-in (no kernel modules) using opensource broadcom drivers (nl80211)
• KSM=y (ROM needs to enable KSM in dalvik and set scan in the init file like in CM9)
• Overclock - needs init.d to be functional to run a init.d script on boot (see sample script below).
• Same system modes apply as before (system_mode=3 for 1.6GHz, system_mode=4 for 1.7GHz, system_mode=5 for 1.8GHz etc.)
• If you don't get deep sleep, make sure you have implemented the init.d script properly. "echo N > /sys/module/cpu_tegra3/parameters/no_lp" seems to be necessary on generic ROMs to get the proper deep sleep.

• Added Simple I/O Scheduler (SIO) - by popular demand.
• deadline i/o scheduler - setting tweaks back to stock. Starving a second write is probably best for performance.
• block: block-exec: blk_execute_rq() - change hang_check timer code back to stock Linux/Nvidia (now also matches the HTC OneX kernel)
• MMC - removed Asus debug printk spamming during normal operations
• MMC - CONFIG_ZIMMER by HTC Studio Engineering (reference OneX kernel) - Experimental - can't find the source of this hack, so it may be an HTC original or a variant of another hack upstream. It increases vm swappiness to 100 by default to force swapping of dead RAM. The theory is to clear out the stale RAM and open up the RAM for something more useful. While this may be counter-intuitive to I/O performance since swapping equates to more disk i/o, the code changes also implement some changes to the block, fs, and mm portions of the kernel. Essentially, from what I can tell, it flags and handles disk operations (aka "bios") differently if they are involved in swap operations (going to or coming from swap). A "bio" is essentially a manifest of an ongoing I/O block device operation, what sectors and memory locations it contains etc. Once the RAM is freed up, I have also been increasing the Dalvik vm RAM settings in the build.prop (see Demandarin's build.prop post) to allow for more RAM consumption. So far, this combination (at least for me so far) has really given me good UI performance and I haven't had to kill the browser often like I did before.
• MMC - increase command retries to 10 from 3 (Nvidia)
• ARM: vfp: only clear vfp state for current cpu in vfp_pm_suspend (github)

• Merged Asus 9.4.2.28 kernel changes (GPS etc. See github for details)
• Removed CONFIG_ASUS_OVERCLOCK_MODES since I don't use it anymore

• Userspace voltage control/undervolting (UV)- experimental, don't set on boot until you test your settings (use SetCPU or other voltage tweaking tool) (thanks richardtrip)
• ZRAM enabled – experimental, needs init.d script "91zram" to enable 100MB zram disk (see below).
• Updated SHA-1 implementation to hopefully improve boot performance
• Memcopy enhancements
• Nvidia patches:
• tps80031 regulator calculation fixes.
• Spi fixes
• Allow CPU-supported unaligned accesses - may provide slight performance improvement
• Improvement for solid state drives.
• Turned off GENTLE_FAIR_SLEEPERS - Theoretically should improve UI performance on Android.
• kernel/mutex.c: update to upstream 3.0 - this fixes various mutex_lock problems.
• Recommend using 1.7GHz if you are looking for Overclock. 1.8GHz is still present, but the TFP does run quite hot and it is kind of overkill in my opinion.
• Updated ARM toolchain to Code Sourcery 2010.09-51

#!/system/bin/sh 
echo 4 > /sys/module/cpu_tegra/parameters/system_mode
echo 1350 > /sys/kernel/tegra_cap/core_cap_level
echo 0 > /sys/kernel/tegra_cap/core_cap_state
echo 0 > /sys/devices/platform/tegra_grhost/enable_3d_scaling
echo 10 > /sys/module/cpu_tegra3/parameters/mp_overhead
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_1
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_2
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_3
echo 9999999 > /sys/module/cpu_tegra/parameters/pwr_cap_limit_4
setprop power.saving.mode 1
setprop persist.tegra.NV_FPSLIMIT 0
setprop persist.sys.NV_FPSLIMIT 0
setprop persist.sys.NV_POWERMODE 0

echo 1700000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
echo N > /sys/module/cpu_tegra3/parameters/no_lp

#!/system/bin/sh
# auto zram activation init script with busybox search
# by show-p1984

echo "[90ZRAM]: Firing up /system/etc/init.d/90zram";

if [ ! -e /sys/block/zram0/disksize ] ; then
        echo "[90ZRAM]: ERROR unable to find /sys/block/zram0/disksize";
        echo "[90ZRAM]: Is this a ZRAM kernel?";
        echo "[90ZRAM]: ZRAM NOT ACTIVATED. (404)";
else
        #find busybox in /system
        bblocation=$(find /system/ -name 'busybox')
        if [ -n "$bblocation" ] && [ -e "$bblocation" ] ; then
                echo "[90ZRAM]: busybox found in:" $bblocation;
                echo "[90ZRAM]: Setting ZRAM disksize.";
                echo $((100*1024*1024)) > /sys/block/zram0/disksize

                echo "[90ZRAM]: Starting ZRAM...";
                bblocation=${bblocation%/*}
                cd $bblocation
                ./busybox mkswap /dev/block/zram0
                ./busybox swapon /dev/block/zram0

                echo "[90ZRAM]: ZRAM activated.";
        else
                echo "[90ZRAM]: ERROR! busybox not found!";
                echo "[90ZRAM]: Is busybox installed? Symlinks set?";
                echo "[90ZRAM]: ZRAM NOT ACTIVATED. (404)";
        fi
fi

Loccy said:

Has Motley forsaken us for the Nexus? Loads of cool stuff going on over there but the Prime kernel scene seems dead as a doornail.

Click to expand...

Click to collapse

st33med said:

Thanks for making another kernel mr motley!

The kernel is working wonders with Paranoid Android on my 64 GB Prime. Have you implemented any of the optimizations from your old kernel for eMMC? I don't know if the driver for it is updated or not.

Typing is a dream now since there is almost no lag now.

Click to expand...

Click to collapse

Hawke84 said:

Hi Motley, can i ask would it be possible for you to include SIO IO scheduler in your kernel? we had it on ICS and it was faster than others, well felt like anyway!

thanks

Click to expand...

Click to collapse

st33med said:

The ROW scheduler is now default. It aids in the slow flash memory, especially ours, by prioritizing Read Over Write, which is what the acronym represents. ROW is nearly identical to SIO except what I said above. This makes it so that fsync writes are scheduled further back and do not interfere with main operation all the time. It still does not solve the problem presented with our slow eMMC, but it helps loads.

Click to expand...

Click to collapse

Hawke84 said:

Thanks, is there a motley kernel for stock JB base 4.1? I couldnt find one! Thanks

Sent from my Transformer Prime TF201 using XDA Premium HD app

Click to expand...

Click to collapse

Bastaerd said:

Latest beta works on stock Asus ROM but it breaks wifi.

Click to expand...

Click to collapse