[Q] How to run blktrace on Android - Android Q&A, Help & Troubleshooting

Hi,
I am trying to run blktrace on my Samsung Galaxy S3; and I compiled my own Android ROM from source code found here: git://github.com/Root-Box/platform_manifest.git (it is Android 4.1.2). I have changed certain configurations so that blktrace is included in the ROM. The linux kernel I am using is 3.0.31-CM
However, when ever I try to run blktrace, I got the following error:
BLKTRACESETUP(2) /dev/block/mmcblk0p1 failed: 25/Not a typewriter.
After using strace to trace the system call, I found that ioctl on /dev/block/mmcblk0p1 with cmd BLKTRACESETUP failed, however I couldn't figure out why.
So I am wondering, did you guys successfully run blktrace on Android; and if so, how did you do that? Also, does anyone has any thoughts on what might went run in my situation?
Thanks a lot.
Suli

So I finally figure this out:
1. In order to use blktrace, the kernel has to be compiled with block tracing support
kernel hacking => tracer => enabling IO tracer
CONFIG_BLKDEV_IO_TRACE
I also enabled a few other tracing options just to be safe
2. In order to use blktrace, all cores have to be online
I observed that blktrace could do ioctl for cpu0 successfully, but not for other cores.
Kernel blktrace facility is creating the trace[n] file in sysfs for the n-th online cpu core which blktrace tries to open.
Usually, only cpu0 is online and others are offline.
That's why blktrace failed opening trace[1, 2, 3] and exited.
So cpufreq governor needs to be changed into "performance" mode
After these two tweaks, it is working

yangsuli said:
So I finally figure this out:
1. In order to use blktrace, the kernel has to be compiled with block tracing support
kernel hacking => tracer => enabling IO tracer
CONFIG_BLKDEV_IO_TRACE
I also enabled a few other tracing options just to be safe
2. In order to use blktrace, all cores have to be online
I observed that blktrace could do ioctl for cpu0 successfully, but not for other cores.
Kernel blktrace facility is creating the trace[n] file in sysfs for the n-th online cpu core which blktrace tries to open.
Usually, only cpu0 is online and others are offline.
That's why blktrace failed opening trace[1, 2, 3] and exited.
So cpufreq governor needs to be changed into "performance" mode
After these two tweaks, it is working
Click to expand...
Click to collapse
I tried to use echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor, however, I still got the error on my nexus 4:
[email protected]:/ # blktrace -d /dev/block/mmcblk0p23 -o sdcard/
Thread 3 failed open /sys/kernel/debug/block/mmcblk0p23/trace3: 2/No such file or directory
Thread 2 failed open /sys/kernel/debug/block/mmcblk0p23/trace2: 2/No such file or directory
Thread 1 failed open /sys/kernel/debug/block/mmcblk0p23/trace1: 2/No such file or directory
FAILED to start thread on CPU 1: 1/Operation not permitted
FAILED to start thread on CPU 2: 1/Operation not permitted
FAILED to start thread on CPU 3: 1/Operation not permitted

request for help
yangsuli said:
So I finally figure this out:
1. In order to use blktrace, the kernel has to be compiled with block tracing support
kernel hacking => tracer => enabling IO tracer
CONFIG_BLKDEV_IO_TRACE
I also enabled a few other tracing options just to be safe
2. In order to use blktrace, all cores have to be online
I observed that blktrace could do ioctl for cpu0 successfully, but not for other cores.
Kernel blktrace facility is creating the trace[n] file in sysfs for the n-th online cpu core which blktrace tries to open.
Usually, only cpu0 is online and others are offline.
That's why blktrace failed opening trace[1, 2, 3] and exited.
So cpufreq governor needs to be changed into "performance" mode
After these two tweaks, it is working
Click to expand...
Click to collapse
Could you tell me how to make cpufreq governor changed into performance mode on nexus

yangsuli said:
Hi,
I am trying to run blktrace on my Samsung Galaxy S3; and I compiled my own Android ROM from source code found here: git://github.com/Root-Box/platform_manifest.git (it is Android 4.1.2). I have changed certain configurations so that blktrace is included in the ROM. The linux kernel I am using is 3.0.31-CM
However, when ever I try to run blktrace, I got the following error:
BLKTRACESETUP(2) /dev/block/mmcblk0p1 failed: 25/Not a typewriter.
After using strace to trace the system call, I found that ioctl on /dev/block/mmcblk0p1 with cmd BLKTRACESETUP failed, however I couldn't figure out why.
So I am wondering, did you guys successfully run blktrace on Android; and if so, how did you do that? Also, does anyone has any thoughts on what might went run in my situation?
Thanks a lot.
Suli
Click to expand...
Click to collapse
Hi, Yangsuli,
For the first step, how to enable I/O tracer please? Just add CONFIG_BLKDEV_IO_TRACE=y in .config? WHat else did you add to the config file?
And Could you tell me how to make cpufreq governor changed into performance mode?
Many thanks!

Related

[KERNEL] Bricked-Kernel Grouper/Tilapia

Welcome to the most customizable N7 kernel on xda
Bricked-Kernel Nexus 7 (grouper)
Replaced NVIDIA's strange hotplug manager with my own: tegra_mpdecision​
Features:
Based upon Googles tegra3 source
Fixed section mismatches
Various other fixes (look @ github)
Compiled with gcc4.7.2 toolchain (linaro 09.12)
Replaced NVIDIAs hotplug manager with tegra_mpdecision (better battery life + performance)
Extensive sysfs interface for mpdecision with all the tuneables you want
Highly modified ondemand governor for tegra3 & mpdecision (this is the only recommended governor atm)
Fixed usage of 51Mhz clock
GPU OC (416/484/520Mhz)
Truly variant free CPU OC for all devices (51 - 1600Mhz)
Fixed max cpufreq resets throughout the kernel
All cores now use the max frequency (before: only in singlecore, otherwise -100Mhz)
Undervolting (faux123)
Default clocks: 102min & 1300max (+100Mhz overclocked) & 475 Maxscroff
earlysuspend/lateresume for ondemand including special screen off settings to save power
Aroma Installer by amarullz
cmdline interface for cpu min/max/maxscroff/gpuoc/scheduler/governor
---Just flash the kernel with your desired settings, no more need for oc apps. Reflash if you want to change sth. It's that easy!
FSync Control by Ezekeel
export t3 variant to debugfs by faux123 (adb shell cat /sys/kernel/debug/t3_variant)
lowered panel min brightness to 8
ZRAM
enabled NFSv3&4
Check the compare links for the rest ​
Where is tha Changelog???
There will be no more changelogs.
Instead the download pages were outfitted with compare links to github for each download.​
How to install?
Flash through recovery. #done.​
Where to complain about errors/bugs?
Please use the Issuetracker for bugs/errors/feature wishes!
Issuetracker @ https://code.google.com/p/bricked/issues/entry
[email protected]
IRC Chat: Freenode IRC #bricked​
Download:
No Guarantees! If it kills your grandmother or your device, I am NOT responsible! If you understand this:
(If you download, please hit Thanks below my post! Thank you!)
>>> DOWNLOAD <<<​
Donor List:​
> Hall of fame <
With special thanks to all N7 piggy bank donors!
Thank you very much!​
Source:
​
Benchmark:
Knowledge base:
What is tegra_mpdecision?
100% kernel based multi core decision! (should cpu1/2/3 be online or not, do we need the low-power/ninja core now?)
This replaces the stock nvidia hotplug solution.
Check /sys/kernel/tegra_mpdecision/conf/ for the configuration.
startdelay = time until mpdecision starts doing it's magic (20000)
delay = time between checks (130)
pause = if something else plugs in the cpu, fall asleep for 10000ms (10 secs)
scroff_single_core = Force low-power mode if screen is off (1)
enabled = enable(1) or disable(0) mpdecision. This does not affect scroff_single_core!
min_cpus = min cpus to be online, cannot be < 1. Default: 1
max_cpus = max cpus to be online, cannot be > 4. (if you set it to 2 and min_cpus to 1 you will basically have a dualcore) Default: 4
idle_freq = a value against that will be checked if a core +/- is requested. (475000)
lpcpu_up_hsyteresis = minimum number of requests that have to be send in order to switch to low-power mode (4)
lpcpu_down_hsyteresis = minimum number of requests that have to be send in order to switch to g-mode (quadcore-mode). Overridden by a requested freq that is higher than the max supported lp-mode frequency. (2)
If cpu0 is below that value and a core up of another cpu is requested, nothing will happen.
If any other cpu is above that value and a core down of that cpu is requested, nothing will happen. (otherwise it would now put down that cpu even though it is still working, which isn't what we want)
Hot plug thresholds (aka now it gets 'complicated')
This small formula calculates which value will be used: (number_of_cpus_online - 1) * 2
The result of this formula will be the nwns_threshold where a new cpu is hotplugged.
The result of this formula + 1 will be the nwns_threshold where a cpu is unplugged.
nwns_threshold_x = runqueue threshold, if this is reached cpuX will be hot/unplugged
twts_threshold_x = time threshold, this amount of time must have passed for the related action to be taken (hot/unplug)
Example:
One cpu is online.
(1 - 1) * 2 = 0 ergo:
nwns_threshold_0 = cpu1 will be hotplugged at this value
((1 - 1) * 2) + 1 = 1
nwns_threshold_1 = cpu0 will be unplugged at this value
This will regulate when we switch to lp-mode, effectively replacing the quadcore g-mode core with the ninja lp-mode core.
Two cpus are online.
(2 - 1) * 2 = 2 ergo:
nwns_threshold_2 = cpu2 will be hotplugged at this value
((2 - 1) * 2) + 1 = 3
nwns_threshold_3 = cpu1 will be unplugged at this value
etc...
(all times are in ms)
If you want to see the mpdecision magic happening:
Code:
adb shell
cat /proc/kmsg | grep 'MPDEC'
mpdecision's input event boost, aka project butter
This will boost your min cpu speed if you touch the screen or press a button and gives you full control.
In those events the min cpu freq will be risen to a predefined value (look below) on every online cpu. This boosts overall reaction times and smoothness a lot. (works similar to the qcom mpdecision binary)
Configuration files:
[email protected]:/sys/kernel/tegra_mpdecision/conf # ls
boost_enabled
boost_freqs
boost_time
All of them work like the usual sysfs files, except one special case:
boost_freqs will list all frequencies from cpu 0 to cpu x. Cpu 3 and any following cpu will share one frequency.
To change those frequencies echo the cpu number + the frequency in khz. To change the boost freq of cpu3 (and 4,5,6,7,8, etc) the echo would look as follows:
Code:
echo "3 1026000" > /sys/kernel/tegra_mpdecision/conf/boost_freqs
for cpu0:
Code:
echo "0 1026000" > /sys/kernel/tegra_mpdecision/conf/boost_freqs
Defaults:
Code:
cat /sys/kernel/tegra_mpdecision/conf/boost_freqs
910000
910000
760000
620000
Recommended governor?
Ondemand.​Why do I have no WLAN?
This kernel is too big for our boot.img with WLAN included into the kernel, so it is built as a module. That means it needs to be inserted into the kernel upon boot up, which needs to be automated for maximum comfort = userinit (init.d) support.
The zip adds init.d support to your ramdisk, if that fails for some reason the wlan module cannot be inserted.
if
Code:
adb shell lsmod
doesn't show this:
Code:
tun 14060 0 - Live 0x00000000
cifs 239646 0 - Live 0x00000000
bcmdhd 433083 0 - Live 0x00000000
Then init.d support is probably missing.
You can test it by going to /system/etc/init.d/ and executing 00wlan.
Code:
adb shell
su
. /system/etc/init.d/00wlan
Wait a few seconds and try to enable wifi and repeat the above lsmod command (should now show bcmdhd as live, you can execute 01cifs and 02tun too).
If WLAN now works, init.d support is missing.
You need to have busybox installed in order to be able to get init.d support going, this shouldn't be a problem for most people, since it is usually installed when rooting your phone anyway (and who does unlock but not root their phone? :/)
It does not matter if busybox is installed in /system/bin or /system/xbin, my script will detect the location upon flashing and patch the ramdisk accordingly.
If no busybox is found, the script will add init.d support to your ramdisk with the default location: /system/xbin/busybox. All you need to do is install busybox there and reboot your phone.
​
show-p1984 said:
Knowledge base:
(under construction)
tegra_mpdecision manual:
coming soon(tm)
Click to expand...
Click to collapse
looks great is this kernal compatible with cm10 based roms such as paranoid android
Flypants101010 said:
looks great is this kernal compatible with cm10 based roms such as paranoid android
Click to expand...
Click to collapse
Yes, ofc
Running cm10 nightlies myself.
show-p1984 said:
Yes, ofc
Running cm10 nightlies myself.
Click to expand...
Click to collapse
All AOSP roms work with Bricked
* Hello World *
What tweaks did you make to on-demand, if you don't mind me asking? The only thing I've noticed is that frequency switching seems to happen faster & more often.
Feel like flashing something tonight!
Sent from my Nexus 7 using xda premium
I tried flashing this a few times and it fails with each try.
Dabaum said:
What tweaks did you make to on-demand, if you don't mind me asking? The only thing I've noticed is that frequency switching seems to happen faster & more often.
Click to expand...
Click to collapse
I mostly adapted it to work with the lpcore while the screen is on and also with special power saving setting while the screen is off.
The 2 phase frequency was changed to accomplish that.
It also applies a powersave bias while screen is off.
Additionally I changed the settings to allow faster scaling with less load. (70% > freq incr)
Source is currently being uploaded, but since I kept the history intact it is taking ages. You can look it up there in more detail once the source is up
airmaxx23 said:
I tried flashing this a few times and it fails with each try.
Click to expand...
Click to collapse
With what message?
I guess because your recovery does not identify your device as 'grouper'.
show-p1984 said:
With what message?
I guess because your recovery does not identify your device as 'grouper'.
Click to expand...
Click to collapse
It just says failed, I've had no problems flashing anything else and I'm flashing mutilple times a day usually.
airmaxx23 said:
It just says failed, I've had no problems flashing anything else and I'm flashing mutilple times a day usually.
Click to expand...
Click to collapse
It most certainly does not say "failed" and nothing else. Look harder.
Using twrp recovery me and my beta testee had no problems at all.
show-p1984 said:
It most certainly does not say "failed" and nothing else. Look harder.
Using twrp recovery me and my beta testee had no problems at all.
Click to expand...
Click to collapse
It says "unable to open zip file. Error flashing zip."
show-p1984 said:
Yes, ofc
Running cm10 nightlies myself.
Click to expand...
Click to collapse
And just to be clear, with non-CM based ones as well? I'm running the 'Glazed' ROM....
airmaxx23 said:
It just says failed, I've had no problems flashing anything else and I'm flashing mutilple times a day usually.
Click to expand...
Click to collapse
Nothing wrong with the zipfile, maybe corrupted download? did you do md5 hash check before flash?
It's either your custom recovery or a corrupted download, period.
ps. Happy Flashing.
airmaxx23 said:
It says "unable to open zip file. Error flashing zip."
Click to expand...
Click to collapse
corrupt download? checked md5/sha1?
corrupt phone upload? checked md5/sha1?
friedsonjm said:
And just to be clear, with non-CM based ones as well? I'm running the 'Glazed' ROM....
Click to expand...
Click to collapse
As long as it is AOSP you should be fine.
But just try it and report back, then we are all a bit more enlightened
(Sorry, I can't possibly test all the ROMs out there ^^)
T-junk said:
Nothing wrong with the zipfile, maybe corrupted download? did you do md5 hash check before flash?
It's either your custom recovery or a corrupted download, period.
ps. Happy Flashing.
Click to expand...
Click to collapse
If I try downloading it directly to my tablet the resulting file is only 17.51K in size, downloading it with a computer takes care of the problem.
airmaxx23 said:
It says "unable to open zip file. Error flashing zip."
Click to expand...
Click to collapse
I get the same.
Sent from my Nexus 7 using xda premium
airmaxx23 said:
If I try downloading it directly to my tablet the resulting file is only 17.51K in size, downloading it with a computer takes care of the problem.
Click to expand...
Click to collapse
StrangerWeather said:
I get the same.
Click to expand...
Click to collapse
I guess same problem for you StangerWeather.
Try downloading on your pc, I will look into that webserver issue tomorrow. (or maybe it's a chrome issue? who knows ^^)

Instant crash when governor set to msm-dcvs

I just got my GSIV today and rooted it using the stickied root method. I have installed setcpu. The rom crashes the second I set the governor to MSM-DCVS. Any ideas?
PelPix said:
I just got my GSIV today and rooted it using the stickied root method. I have installed setcpu. The rom crashes the second I set the governor to MSM-DCVS. Any ideas?
Click to expand...
Click to collapse
My S4 GT-i9505 crash too...
PelPix said:
I just got my GSIV today and rooted it using the stickied root method. I have installed setcpu. The rom crashes the second I set the governor to MSM-DCVS. Any ideas?
Click to expand...
Click to collapse
Same here.... i searched and really have not found an answer.
Stacktrace from enabling msm-dcvs
Here's the relevant stack trace from the kernel panic when enabling the msm-dcvs governor.
Code:
[ 2811.452880] MSM_DCVS: msm_dcvs_freq_sink_start: msm_dcvs_freq_sink_start invalid dcvs_core_id = 0 returning -EINVAL
[ 2811.462585] ------------[ cut here ]------------
[ 2811.466979] kernel BUG at /opt/jenkins/android/workspace/android/jellybean/kernel/samsung/jf/drivers/cpufreq/cpufreq_gov_msm.c:206!
[ 2811.478790] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
[ 2811.484557] Modules linked in: dhd vpnclient
[ 2811.488799] CPU: 0 Not tainted (3.4.0-cyanogenmod-gfbaa5c7 #1)
I've done a little cursory exploring and noticed that the CPU_OFFSET parameter starts at 1 on this kernel module, but it is being told to enable this governor on core 0, hence the message above. Not sure at the moment if it's the module's fault, or just some misconfiguration from settings activity when enabling this module.
I plan to look into this in a bit more depth sometime in the next few weeks(hopefully).
Follow up to stepheno's message
stephenopdx said:
Here's the relevant stack trace from the kernel panic when enabling the msm-dcvs governor.
Code:
[ 2811.452880] MSM_DCVS: msm_dcvs_freq_sink_start: msm_dcvs_freq_sink_start invalid dcvs_core_id = 0 returning -EINVAL
[ 2811.462585] ------------[ cut here ]------------
[ 2811.466979] kernel BUG at /opt/jenkins/android/workspace/android/jellybean/kernel/samsung/jf/drivers/cpufreq/cpufreq_gov_msm.c:206!
[ 2811.478790] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
[ 2811.484557] Modules linked in: dhd vpnclient
[ 2811.488799] CPU: 0 Not tainted (3.4.0-cyanogenmod-gfbaa5c7 #1)
I've done a little cursory exploring and noticed that the CPU_OFFSET parameter starts at 1 on this kernel module, but it is being told to enable this governor on core 0, hence the message above. Not sure at the moment if it's the module's fault, or just some misconfiguration from settings activity when enabling this module.
I plan to look into this in a bit more depth sometime in the next few weeks(hopefully).
Click to expand...
Click to collapse
Hi, stephenopdx and I are working together on this issue in our free time. As a follow up, we've found that the call to msm_dcvs_scm_init is returning a non-zero value (specifically 4). This function is calling the command DCVD_CMD_INIT (command ID 5) in the SCM_SVC_DCVS service. The continues down to an ARM instruction call to move the processor into "secure mode" (aka TrustZone). I'd like to emphasize at this point that both stephenopdx are complete noobs when it comes to kernel development.
We are wondering:
1) Why, if the scm_call to DCVS_CMD_INIT fails, does the kernel continue on its merry way?
2) What does that return code of 4 mean?
3) What preconditions are necessary to initialize DCVS?
4) Is there a place where we can find reference materials (things like return codes and their meanings) for the Qualcomm MSM 8960 SoC? Googling for return code 4 was fairly futile.
5) Is there some sort of kernel debugger so that we can stop using terrible pr_err messages?
We can also produce stack traces or kernel panic logs (heavily augmented with demented pr_err messages to help us figure out what's going on) if anyone is interested.
Thanks!

[Q] Default settings of the ondemand CPUFreq governor parameters in SAMSUNG S4

I would like to know the default settings of the parameters of ondemand CPUFreq governor that SAMSUNG has choosen for my SAMSUNG S4. I have not flashed my phone with any custom kernel. I have been reading about dynamic frequency scaling support in Linux Kernel and found out that CPUFreq subsystem implements this functionality with the help of the governors.
I found the governor for the Core 0 of my device as
:~$ adb shell cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
ondemand
then I started learning about ondemand governor. I found in kernel documentation, which states that following are the parameters for ondemand governor.
•sampling_rate
•sampling_rate_min
•up_threshold
•ignore_nice_load
•sampling_down_factor
•powersave_bias
next, I was trying to figure out where/how can I access these parameters, I read here
and some other places as well that these parameters are found in /sys/devices/system/cpu/cpu?/cpufreq/ondemand directory.
Now, In my case for SAMSUNG S4, I can not see any directory named as ondemand inside /sys/devices/system/cpu/cpu0/cpufreq . I tried a rooted SAMSUNG S3 of a friend aswell, and I could not see the directory in that aswell.
My intention is to keep the default settings/not change the kernel and just to get to know what are the default settings of the CPUFreq governor, I purchased the phone with ?
P.S. this is my first post on xda-developer forums, so please pardon me if I ignored some important rule.

[DEV only] Random Development thoughts

I have started this thread as a place to discuss random development thought that probably don't need their own threads, and as a place to discuss issues so other threads don't go too far off topic.
Table of Contents:
DBV for c6843: Posts 1 - 6
GPL: Post 7
F2FS: Post 8
wlan: post 9, 11 -
Optimisation Bookmark: post 10
Original Post 1:
First thought
@dbolivar Have you tried building you (DooMKernel) with CONFIG_DVB_CORE = y/m to see if you can get the TV function to work?
blueether said:
I have started this thread as a place to discuss random development thought that probably don't need their own threads, and as a place to discuss issues so other threads don't go too far off topic.
First thought
@dbolivar Have you tried building you (DooMKernel) with CONFIG_DVB_CORE = y/m to see if you can get the TV function to work?
Click to expand...
Click to collapse
Hi, good start. In fact I have, the differences in the kernel options between C6833 and C6843 (Brazilian model with digital TV) are:
Code:
47c47
< CONFIG_MACH_SONY_TOGARI_BRAZIL=y
---
> CONFIG_MACH_SONY_TOGARI_ROW=y
275d274
< CONFIG_ISDBT_TUNER_SMTEJ11X=y
325a325,327
> CONFIG_TOUCHSCREEN_CLEARPAD=y
> CONFIG_TOUCHSCREEN_CLEARPAD_I2C=y
> CONFIG_TOUCHSCREEN_CLEARPAD_RMI_DEV=y
Yes, I don't know why the Clearpad touchscreen is enabled for the C6833, but not for the C6843; I think it's just garbage (perhaps they made the togari defconfig based off honami, and corrected that when making the togari_brazil defconfig). It doesn't make any difference for me, disabling them keep the touchscreen (MAX1187) functions normal.
Well, back to the point: even when applying these differences to the ZU DooMKernel, the digital TV doesn't work. The app stays a long time in a black screen, and finally closes (FC). I investigated the kmsg and logcat, but nothing useful.
CONFIG_DVB_CORE is "m" by default. I tried insmod'ing every module in /system/lib/modules, but same result. I temporarily gave up, because I don't watch the crap they broadcast on the open TV here. But yeah, it would be good to have everything working...
dbolivar said:
Hi, good start. In fact I have, the differences in the kernel options between C6833 and C6843 (Brazilian model with digital TV) are:
Code:
47c47
< CONFIG_MACH_SONY_TOGARI_BRAZIL=y
---
> CONFIG_MACH_SONY_TOGARI_ROW=y
275d274
< CONFIG_ISDBT_TUNER_SMTEJ11X=y
325a325,327
> CONFIG_TOUCHSCREEN_CLEARPAD=y
> CONFIG_TOUCHSCREEN_CLEARPAD_I2C=y
> CONFIG_TOUCHSCREEN_CLEARPAD_RMI_DEV=y
Yes, I don't know why the Clearpad touchscreen is enabled for the C6833, but not for the C6843; I think it's just garbage (perhaps they made the togari defconfig based off honami, and corrected that when making the togari_brazil defconfig). It doesn't make any difference for me, disabling them keep the touchscreen (MAX1187) functions normal.
Well, back to the point: even when applying these differences to the ZU DooMKernel, the digital TV doesn't work. The app stays a long time in a black screen, and finally closes (FC). I investigated the kmsg and logcat, but nothing useful.
CONFIG_DVB_CORE is "m" by default. I tried insmod'ing every module in /system/lib/modules, but same result. I temporarily gave up, because I don't watch the crap they broadcast on the open TV here. But yeah, it would be good to have everything working...
Click to expand...
Click to collapse
I hadn't got around to diff'ing them I just noticed the DVB line and had a quick look at the brazil defconfig.
I guess there is a binary blob that is needed?
For some reason I have to include the clearpad stuff at the moment in the pimped kernel, I should dig to find out why so the kernel is a tad smaller.
blueether said:
I hadn't got around to diff'ing them I just noticed the DVB line and had a quick look at the brazil defconfig.
I guess there is a binary blob that is needed?
For some reason I have to include the clearpad stuff at the moment in the pimped kernel, I should dig to find out why so the kernel is a tad smaller.
Click to expand...
Click to collapse
Hmm very good insight about the binary blob... It's very possible. I have FTFs for the C6833 and C6843, I'll extract them and compare the file list.
I made it work!! Well, involves some manual steps, but it's a very good progress. I started by comparing the file listing in /system from stock C6833 and C6843 firmwares, and although there is no kernel blob, there ARE some files related to DTV, mostly dynamic libs (.so), some configuration, but most importantly, these two:
Code:
/system/bin/dtvsdserver
/system/bin/dtvserver
So, I insmod'ed every module related to DTV/DVB, and then started these guys... And the TV works! BUT, as soon as I quit the Mobile TV app, the process "dtvserver" is killed with a hangup signal, and opening the app again does not restart it, so I have to manually bring the dtvserver up first (otherwise the same previous issue happens, black screen and finally Mobile TV FC's).
I notice some DTV/DVB modules couldn't be insmod'ed:
Code:
insmod: init_module 'dib3000mc.ko' failed (No such file or directory)
insmod: init_module 'dib7000m.ko' failed (No such file or directory)
insmod: init_module 'dib7000p.ko' failed (No such file or directory)
insmod: init_module 'dib8000.ko' failed (No such file or directory)
insmod: init_module 'dib9000.ko' failed (No such file or directory)
insmod: init_module 'mpq-dmx-hw-plugin.ko' failed (No such file or directory)
insmod: init_module 'tuner-simple.ko' failed (No such file or directory)
But the modules are there, and they are the new ones... I don't get it why they can't be insmod'ed, perhaps something is missing from the kernel config? It could be one of the reasons.
Another thing I've been wondering, is why the DVB modules necessary for the Mobile TV are not loaded automatically, like the wlan module is, for instance. Perhaps a difference in the ramdisk? Something to investigate next.
dbolivar said:
I made it work!! Well, involves some manual steps...
Another thing I've been wondering, is why the DVB modules necessary for the Mobile TV are not loaded automatically, like the wlan module is, for instance. Perhaps a difference in the ramdisk? Something to investigate next.
Click to expand...
Click to collapse
Congrats on getting that far with the DTV, will make for a more complete device for the c6843 users that want to use custom ROMs
Any differences in udev or init.rc or whatever android uses?
Development using the GPL and GPL'd code
The GPL is a very powerful tool, as are the other OSS licences. One has to know how they are used and can/can't be enforced before publishing any code under these licences or modifying code that that been published under a OSS licence. Using a OSS licence brings with it risks and rewords, often these are one and the same.
When we do anything with Android [Linux] kernels we fall under the bounds of the GPL v2 that it is published under. Any modification to the kernel has to be published under the same licence, and we have to publish the newly modified source in a sate that can recreate the binary that you publish to the public/xda.
I'll leave you with this quote from http://programmers.stackexchange.com
...are you prepared to live by it, and let other people use what you've written, rather than just liking it because of what you can get out of it?
Click to expand...
Click to collapse
F2FS mount and sysfs options
Now that we have an F2FS-enabled firmware by default, I've been reading about the mount options and sysfs entries available. As many of us know, there are lots of tweaks out there for EXT4, but what about F2FS? Well, here is the official documentation from the Linux kernel:
https://www.kernel.org/doc/Documentation/filesystems/f2fs.txt
As for the mount options, I think only discard could be interesting (but it's questionable, some people say it's better to schedule a regular fstrim call instead). There are also some sysfs tunables which I think can have a noticeable impact in battery life, and performance. Testing them would require a lot of trial and error. I'm pasting them below for those who don't want to go through the full document:
Code:
================================================================================
SYSFS ENTRIES
================================================================================
Information about mounted f2f2 file systems can be found in
/sys/fs/f2fs. Each mounted filesystem will have a directory in
/sys/fs/f2fs based on its device name (i.e., /sys/fs/f2fs/sda).
The files in each per-device directory are shown in table below.
Files in /sys/fs/f2fs/<devname>
(see also Documentation/ABI/testing/sysfs-fs-f2fs)
..............................................................................
File Content
gc_max_sleep_time This tuning parameter controls the maximum sleep
time for the garbage collection thread. Time is
in milliseconds.
gc_min_sleep_time This tuning parameter controls the minimum sleep
time for the garbage collection thread. Time is
in milliseconds.
gc_no_gc_sleep_time This tuning parameter controls the default sleep
time for the garbage collection thread. Time is
in milliseconds.
gc_idle This parameter controls the selection of victim
policy for garbage collection. Setting gc_idle = 0
(default) will disable this option. Setting
gc_idle = 1 will select the Cost Benefit approach
& setting gc_idle = 2 will select the greedy aproach.
reclaim_segments This parameter controls the number of prefree
segments to be reclaimed. If the number of prefree
segments is larger than the number of segments
in the proportion to the percentage over total
volume size, f2fs tries to conduct checkpoint to
reclaim the prefree segments to free segments.
By default, 5% over total # of segments.
max_small_discards This parameter controls the number of discard
commands that consist small blocks less than 2MB.
The candidates to be discarded are cached until
checkpoint is triggered, and issued during the
checkpoint. By default, it is disabled with 0.
ipu_policy This parameter controls the policy of in-place
updates in f2fs. There are five policies:
0: F2FS_IPU_FORCE, 1: F2FS_IPU_SSR,
2: F2FS_IPU_UTIL, 3: F2FS_IPU_SSR_UTIL,
4: F2FS_IPU_DISABLE.
min_ipu_util This parameter controls the threshold to trigger
in-place-updates. The number indicates percentage
of the filesystem utilization, and used by
F2FS_IPU_UTIL and F2FS_IPU_SSR_UTIL policies.
max_victim_search This parameter controls the number of trials to
find a victim segment when conducting SSR and
cleaning operations. The default value is 4096
which covers 8GB block address range.
dir_level This parameter controls the directory level to
support large directory. If a directory has a
number of files, it can reduce the file lookup
latency by increasing this dir_level value.
Otherwise, it needs to decrease this value to
reduce the space overhead. The default value is 0.
ram_thresh This parameter controls the memory footprint used
by free nids and cached nat entries. By default,
10 is set, which indicates 10 MB / 1 GB RAM.
@blueether & @dbolivar :
our primary concern should be to get custom wlan (prima/pronto) drivers working on custom compiled kernel for stock ROMs.. once we have that there are a LOT of features which we can add!
so far I havent had much luck but you can find all my tests here
https://github.com/DooMLoRD/android_kernel_sony_msm8974/commits/testing_stock_4.3_wlan
https://github.com/DooMLoRD/android_kernel_sony_msm8974/commits/testing_stock_4.3_wlan_clean
https://github.com/DooMLoRD/android_kernel_sony_msm8974/commits/testing_z1_gpe_port_wlan
A bookmark
Worth the read @dbolivar
http://forum.xda-developers.com/showthread.php?t=2754997
DooMLoRD said:
@blueether & @dbolivar :
our primary concern should be to get custom wlan (prima/pronto) drivers working on custom compiled kernel for stock ROMs.. once we have that there are a LOT of features which we can add!
so far I havent had much luck but you can find all my tests here
https://github.com/DooMLoRD/android_kernel_sony_msm8974/commits/testing_stock_4.3_wlan
https://github.com/DooMLoRD/android_kernel_sony_msm8974/commits/testing_stock_4.3_wlan_clean
https://github.com/DooMLoRD/android_kernel_sony_msm8974/commits/testing_z1_gpe_port_wlan
Click to expand...
Click to collapse
I'm giving it a try - downloaded the latest sources from CodeAurora and applied to your kernel (adjusting Kconfigs, Makefiles and defconfig). The compilation stops here:
Code:
drivers/staging/prima/CORE/HDD/src/wlan_hdd_cfg80211.c: In function ‘wlan_hdd_send_avoid_freq_event’:
drivers/staging/prima/CORE/HDD/src/wlan_hdd_cfg80211.c:574:18: warning: assignment makes pointer from integer without a cast
error, forbidden warning: wlan_hdd_cfg80211.c:574
make[3]: *** [drivers/staging/prima/CORE/HDD/src/wlan_hdd_cfg80211.o] Error 1
OK, so I edited wlan_hdd_cfg80211.c and changed line 574 to add an explicit cast:
Code:
vendor_event = (struct sk_buff *)cfg80211_vendor_event_alloc(pHddCtx->wiphy,
Then it passes by this point, but stops in following lines with similar errors (not even pasting here). What I noticed from these errors it that CodeAurora's sources seem to be incompatible with Sony's kernel, because for instance, the function cfg80211_vendor_event_alloc is not defined anywhere (ran a grep over all *.c and *.h files in the kernel).
Comparing with a CM11 kernel, this function is also not defined anywhere, but the prima sources are very different, they don't call it. So I thought: let's use CM11's prima sources.
Bingo!
Compiled fine, booted, and with the correspoding firmware in /system/etc/firmware/wlan/prima, WLAN came up normally after boot. :victory:
Now you mentioned something about CM11's prima sources, from what I understood they are not ideal. Is that true? Because there are CM11-based custom kernels which implement intelliplug (dependent on custom WLAN drivers), for instance.
Hope it works for you. In this case, I'll put together a more organized how-to of what I did.
dbolivar said:
I'm giving it a try - downloaded the latest sources from CodeAurora and applied to your kernel (adjusting Kconfigs, Makefiles and defconfig). The compilation stops here:
Code:
drivers/staging/prima/CORE/HDD/src/wlan_hdd_cfg80211.c: In function ‘wlan_hdd_send_avoid_freq_event’:
drivers/staging/prima/CORE/HDD/src/wlan_hdd_cfg80211.c:574:18: warning: assignment makes pointer from integer without a cast
error, forbidden warning: wlan_hdd_cfg80211.c:574
make[3]: *** [drivers/staging/prima/CORE/HDD/src/wlan_hdd_cfg80211.o] Error 1
OK, so I edited wlan_hdd_cfg80211.c and changed line 574 to add an explicit cast:
Code:
vendor_event = (struct sk_buff *)cfg80211_vendor_event_alloc(pHddCtx->wiphy,
Then it passes by this point, but stops in following lines with similar errors (not even pasting here). What I noticed from these errors it that CodeAurora's sources seem to be incompatible with Sony's kernel, because for instance, the function cfg80211_vendor_event_alloc is not defined anywhere (ran a grep over all *.c and *.h files in the kernel).
Comparing with a CM11 kernel, this function is also not defined anywhere, but the prima sources are very different, they don't call it. So I thought: let's use CM11's prima sources.
Bingo!
Compiled fine, booted, and with the correspoding firmware in /system/etc/firmware/wlan/prima, WLAN came up normally after boot. :victory:
Now you mentioned something about CM11's prima sources, from what I understood they are not ideal. Is that true? Because there are CM11-based custom kernels which implement intelliplug (dependent on custom WLAN drivers), for instance.
Hope it works for you. In this case, I'll put together a more organized how-to of what I did.
Click to expand...
Click to collapse
great!
can you just fork my repo and shift to the branch which worked for you, then send pull request? i will investigate on the Z1...
and yes please do send the steps!
well if we get custom compiled WLAN modules properly working on custom compiled kernel for stock ROM then i can have features like intelli-plug, F2FS, AsyncFS and lots more on stock ROMs!
DooMLoRD said:
great!
can you just fork my repo and shift to the branch which worked for you, then send pull request? i will investigate on the Z1...
and yes please do send the steps!
well if we get custom compiled WLAN modules properly working on custom compiled kernel for stock ROM then i can have features like intelli-plug, F2FS, AsyncFS and lots more on stock ROMs!
Click to expand...
Click to collapse
Well I'm still learning my way around git & GitHub, so I think it will be faster if you just reproduce the steps below:
1) Copy Prima WLAN sources from CM11-based kernel:
I copied from SlimRom's kernel source. Put the sources under drivers/staging/prima.
2) Modify related Kconfig and Makefile:
drivers/staging/Kconfig
Code:
131,132d130
< source "drivers/staging/prima/Kconfig"
<
drivers/staging/Makefile
Code:
58,59d57
< obj-$(CONFIG_PRIMA_WLAN) += prima/
< obj-$(CONFIG_PRONTO_WLAN) += prima/
3) Modify defconfig:
I'm pasting every option related to WLAN. The Prima options should appear after a "make menuconfig".
Code:
CONFIG_WIRELESS=y
CONFIG_WIRELESS_EXT=y
CONFIG_WEXT_CORE=y
CONFIG_WEXT_PROC=y
CONFIG_WEXT_SPY=y
CONFIG_WEXT_PRIV=y
CONFIG_CFG80211=y
CONFIG_NL80211_TESTMODE=y
# CONFIG_CFG80211_DEVELOPER_WARNINGS is not set
# CONFIG_CFG80211_REG_DEBUG is not set
CONFIG_CFG80211_DEFAULT_PS=y
# CONFIG_CFG80211_DEBUGFS is not set
CONFIG_CFG80211_INTERNAL_REGDB=y
# CONFIG_CFG80211_WEXT is not set
# CONFIG_WIRELESS_EXT_SYSFS is not set
# CONFIG_LIB80211 is not set
# CONFIG_CFG80211_ALLOW_RECONNECT is not set
# CONFIG_MAC80211 is not set
# CONFIG_WIMAX is not set
CONFIG_RFKILL=y
CONFIG_RFKILL_PM=y
CONFIG_RFKILL_LEDS=y
# CONFIG_RFKILL_INPUT is not set
# CONFIG_RFKILL_REGULATOR is not set
# CONFIG_RFKILL_GPIO is not set
...
# CONFIG_WIFI_CONTROL_FUNC is not set
...
CONFIG_WLAN=y
# CONFIG_USB_ZD1201 is not set
# CONFIG_USB_NET_RNDIS_WLAN is not set
# CONFIG_LIBRA_SDIOIF is not set
# CONFIG_ATH6K_LEGACY_EXT is not set
CONFIG_WCNSS_CORE=y
CONFIG_WCNSS_CORE_PRONTO=y
CONFIG_WCNSS_MEM_PRE_ALLOC=y
CONFIG_WCNSS_REGISTER_DUMP_ON_BITE=y
# CONFIG_ATH_COMMON is not set
# CONFIG_BCMDHD is not set
# CONFIG_BRCMFMAC is not set
# CONFIG_HOSTAP is not set
# CONFIG_IWM is not set
# CONFIG_LIBERTAS is not set
# CONFIG_MWIFIEX is not set
...
#
# Qualcomm Atheros Prima WLAN module
#
CONFIG_PRIMA_WLAN=m
CONFIG_PRONTO_WLAN=y
# CONFIG_PRIMA_WLAN_BTAMP is not set
CONFIG_PRIMA_WLAN_LFR=y
CONFIG_PRIMA_WLAN_OKC=y
CONFIG_PRIMA_WLAN_11AC_HIGH_TP=y
CONFIG_WLAN_FEATURE_11W=y
CONFIG_QCOM_VOWIFI_11R=y
CONFIG_CONFIG_ENABLE_LINUX_REG=y
4) Include Prima WLAN firmware in kernel ZIP file:
Copy from a CM11-based firmware (I used PAC-Rom). Remember to set permissions as 644.
/system/etc/firmware/wlan/prima
Code:
WCNSS_cfg.dat
WCNSS_qcom_cfg.ini
WCNSS_qcom_wlan_nv.bin
5) Include wlan.ko in kernel ZIP file:
Actually I include every compiled kernel module, because we are using a different toolchain and they may benefit from it too. Remember that /system/lib/module/wlan.ko is a symlink, but in my test the target got updated accordingly.
dbolivar said:
Well I'm still learning my way around git & GitHub, so I think it will be faster if you just reproduce the steps below:
1) Copy Prima WLAN sources from CM11-based kernel:
I copied from SlimRom's kernel source. Put the sources under drivers/staging/prima.
2) Modify related Kconfig and Makefile:
drivers/staging/Kconfig
Code:
131,132d130
< source "drivers/staging/prima/Kconfig"
<
drivers/staging/Makefile
Code:
58,59d57
< obj-$(CONFIG_PRIMA_WLAN) += prima/
< obj-$(CONFIG_PRONTO_WLAN) += prima/
3) Modify defconfig:
I'm pasting every option related to WLAN. The Prima options should appear after a "make menuconfig".
Code:
CONFIG_WIRELESS=y
CONFIG_WIRELESS_EXT=y
CONFIG_WEXT_CORE=y
CONFIG_WEXT_PROC=y
CONFIG_WEXT_SPY=y
CONFIG_WEXT_PRIV=y
CONFIG_CFG80211=y
CONFIG_NL80211_TESTMODE=y
# CONFIG_CFG80211_DEVELOPER_WARNINGS is not set
# CONFIG_CFG80211_REG_DEBUG is not set
CONFIG_CFG80211_DEFAULT_PS=y
# CONFIG_CFG80211_DEBUGFS is not set
CONFIG_CFG80211_INTERNAL_REGDB=y
# CONFIG_CFG80211_WEXT is not set
# CONFIG_WIRELESS_EXT_SYSFS is not set
# CONFIG_LIB80211 is not set
# CONFIG_CFG80211_ALLOW_RECONNECT is not set
# CONFIG_MAC80211 is not set
# CONFIG_WIMAX is not set
CONFIG_RFKILL=y
CONFIG_RFKILL_PM=y
CONFIG_RFKILL_LEDS=y
# CONFIG_RFKILL_INPUT is not set
# CONFIG_RFKILL_REGULATOR is not set
# CONFIG_RFKILL_GPIO is not set
...
# CONFIG_WIFI_CONTROL_FUNC is not set
...
CONFIG_WLAN=y
# CONFIG_USB_ZD1201 is not set
# CONFIG_USB_NET_RNDIS_WLAN is not set
# CONFIG_LIBRA_SDIOIF is not set
# CONFIG_ATH6K_LEGACY_EXT is not set
CONFIG_WCNSS_CORE=y
CONFIG_WCNSS_CORE_PRONTO=y
CONFIG_WCNSS_MEM_PRE_ALLOC=y
CONFIG_WCNSS_REGISTER_DUMP_ON_BITE=y
# CONFIG_ATH_COMMON is not set
# CONFIG_BCMDHD is not set
# CONFIG_BRCMFMAC is not set
# CONFIG_HOSTAP is not set
# CONFIG_IWM is not set
# CONFIG_LIBERTAS is not set
# CONFIG_MWIFIEX is not set
...
#
# Qualcomm Atheros Prima WLAN module
#
CONFIG_PRIMA_WLAN=m
CONFIG_PRONTO_WLAN=y
# CONFIG_PRIMA_WLAN_BTAMP is not set
CONFIG_PRIMA_WLAN_LFR=y
CONFIG_PRIMA_WLAN_OKC=y
CONFIG_PRIMA_WLAN_11AC_HIGH_TP=y
CONFIG_WLAN_FEATURE_11W=y
CONFIG_QCOM_VOWIFI_11R=y
CONFIG_CONFIG_ENABLE_LINUX_REG=y
4) Include Prima WLAN firmware in kernel ZIP file:
Copy from a CM11-based firmware (I used PAC-Rom). Remember to set permissions as 644.
/system/etc/firmware/wlan/prima
Code:
WCNSS_cfg.dat
WCNSS_qcom_cfg.ini
WCNSS_qcom_wlan_nv.bin
5) Include wlan.ko in kernel ZIP file:
Actually I include every compiled kernel module, because we are using a different toolchain and they may benefit from it too. Remember that /system/lib/module/wlan.ko is a symlink, but in my test the target got updated accordingly.
Click to expand...
Click to collapse
which branch did u test this with?
also can you please upload the correct firmware (/system/etc/firmware/wlan/prima) which worked for you?
DooMLoRD said:
which branch did u test this with?
also can you please upload the correct firmware (/system/etc/firmware/wlan/prima) which worked for you?
Click to expand...
Click to collapse
The branch from your kernel was master_kk-4.4.2. The Prima WLAN sources were, as I mentioned, from SlimRom's kernel, branch kk4.4.
I'm attaching the firmware to this post.
blueether said:
A bookmark
Worth the read @dbolivar
http://forum.xda-developers.com/showthread.php?t=2754997
Click to expand...
Click to collapse
Wow this is very nice! Follows the same line of research I've been doing, but this is more "elegant" and in-depth, especially because of all the benchmarks and other comparisons he has done. I'll try this approach myself on my custom builds.
One thing I noticed is that we may still have to use the "hammer" approach with the find & replace of the "-O" flags in *.mk and Makefiles I've been doing, because some of them have these flags hardcoded (i.e. they don't take the value from core/combo/TARGET_linux-arm.mk or other base files). I'll tip JustArchi in his thread about it.
Hi guys, I posted this in the general android section and have had no answer, maybe some one here might know?
I'm trying to build AOSP using CM 11's device tree for togari/togari_gpe and have hit a brick wall after solving the first few make errors.
I have to errors, that I suspect are the same problem:
The first is if I just build with make I get this error:
Code:
Import includes file: out/target/product/togari/obj/SHARED_LIBRARIES/copybit.msm8974_intermediates/import_includes
Import includes file: out/target/product/togari/obj/SHARED_LIBRARIES/libmemalloc_intermediates/import_includes
target thumb C++: libqdutils <= hardware/qcom/display/msm8974/libqdutils/profiler.cpp
target thumb C++: libqdutils <= hardware/qcom/display/msm8974/libqdutils/mdp_version.cpp
target thumb C++: libqdutils <= hardware/qcom/display/msm8974/libqdutils/idle_invalidator.cpp
target thumb C++: libqdutils <= hardware/qcom/display/msm8974/libqdutils/comptype.cpp
hardware/qcom/display/msm8974/libqdutils/mdp_version.cpp:33:27: fatal error: linux/msm_mdp.h: No such file or directory
#include <linux/msm_mdp.h>
^
Import includes file: out/target/product/togari/obj/SHARED_LIBRARIES/gps.msm8974_intermediates/import_includes
compilation terminated.
make: *** [out/target/product/togari/obj/SHARED_LIBRARIES/libqdutils_intermediates/mdp_version.o] Error 1
make: *** Waiting for unfinished jobs....
and if I do make bootimage I get this:
Code:
make: *** No rule to make target `out/target/product/togari/kernel', needed by `out/target/product/togari/boot.img'. Stop.
I suspect that both stem from aosp not finding the kernel source?
kernel source is at ~/dev/aosp/kernel/sony/msm8974/ and I'm building in ~/dev/aosp
I'm also trying to build vanir aosp using the same device tree and kernel and make bootimage completes fine.
Any help would be great
Yo, @DooMLoRD, @blueether @dbolivar
You guys have probably registred that Geohot (iPhone, PS3 jailbreaker) has successfully rooted the Galaxy S5 (and can now claim the $18,000 bounty), but i just read that this one-click root tool should work on any device running a pre-June 3rd kernel, i.e Android 4.4.2. The question is, does it work on the Ultra? I'm aware that it's easy to root an UItra AS LONG as the bootloader is unlocked, but this thing circumvents that part, and roots it. I think.
Are there anyone here still left on 4.4.2 that can try?
More here: Click
LordManhattan said:
Yo, @DooMLoRD, @blueether @dbolivar
You guys have probably registred that Geohot (iPhone, PS3 jailbreaker) has successfully rooted the Galaxy S5 (and can now claim the $18,000 bounty), but i just read that this one-click root tool should work on any device running a pre-June 3rd kernel, i.e Android 4.4.2. The question is, does it work on the Ultra? I'm aware that it's easy to root an UItra AS LONG as the bootloader is unlocked, but this thing circumvents that part, and roots it. I think.
Are there anyone here still left on 4.4.2 that can try?
More here: Click
Click to expand...
Click to collapse
Can't say... Will have to test
Sent from my C6902 using XDA Free mobile app
DooMLoRD said:
Can't say... Will have to test
Sent from my C6902 using XDA Free mobile app
Click to expand...
Click to collapse
I made a thread here. It should work.
http://forum.xda-developers.com/showthread.php?t=2783982

problem with manually hotplugging cpuX in rooted android device

Hi every one
I have a Lenovo vibe p1a42 which is rooted properly along with the SuperSU.
I'm going to write a custom userspace governor, using cpufreq and hotplug facility but I have some problem with online and some other entries in sysfs.
I have 8 cores and it is possible to change the corresponding entries in sysfs. I can successfully selecting the userspace governor by:
Code:
adb shell
su
echo userspace > /sys/devices/system/cpu/cpuX/cpufreq/scaling_governor
I do the last line for each core and then I set the valid frequency by writing to corresponding scaling_setspeed files.
I know it is possible to dynamically on/off each core by writing 1/0 to the corresponding /sys/devices/system/cpu/cpuX/online entry.
However, I cannot change the status of some processor to online by:
Code:
echo "1" > /sys/devices/system/cpu/cpuX/online
two cores of eight are always off and it seems that it cann't be controlled, manually, in this manner.
I know there is some kind of hotplug manager in android devices which activate appropriate core numbers according to the load.
How can I get ride of such hotplug manager and get full access to all cores due to the maximum performance?

Categories

Resources