Hi,
Altough this is just the skeleton of it, let me share what I've set up until now.
I've also started a dedicated Google group here: https://groups.google.com/forum/?hl=en&fromgroups#!members/renesas-emev-osp
Anyone interested in contributing to develpment, plesae join!
1) Android
https://github.com/ffxx68/Renesas-device_emev
This is the entry-point to the GitHub repositories where I would put all the modification to the AOSP code which are specific to the Renesas EMEV for our Open Source Project.
Have a look at the README for further details and follow the progress, or open discussions, on our Google group.
2) Kernel
https://github.com/ffxx68/RenesasEV2-BSPGB-Kernel
This is the public GitHub repository for the 2.6.35.7 Kernel source code. It is customized to emxx code and it includes for the moment only the LCD patch that is necessary for the most common tablet (the kernel base code is aimed at the development board instead).
Same as above, anyone can contribute directly here.
3) Bootloader
https://github.com/ffxx68/RenesasEV2-bootloader
This is the public GitHub repository for the bootloader source code. It includes a patch for the 4 DDR tablet models.
I don't expect this to be changed much, but anyone can contribute directly pushing new files over here too.
4) Packaging
Some more details about how to wrap-up a fresh build and make a firmware for the tablet are found in the README mentieod above.
I've managed to boot Android from an SD card (see here: https://github.com/ffxx68/Renesas-BootSD) and I think a similar approach should be followed, at least for initial testing.
Kind of late to the party, but I'm really interested in this project... I applied to the Google Group, but I notice that the EV2 AOSP page you listed is gone... is there a specific reason for this? I would really love to contribute to this.
Hi, and welcome aboard!
I have edited a couple links above, which have changed a bit since then.
Thanks
I'm facing a problem with the touchscreen driver, which I have filed in the group list as well:
https://groups.google.com/forum/#!topic/renesas-emev-osp/smcwddwvNDo
(answer over there, if possible).
I can't understand why the driver for the touchscreen (found in drivers/input/touchscreen/pixcir_i2c_ts.c) is not loaded during the kernel init stages.
The touchkeys driver (ekt2201) is very similar in code structure and it's loaded as expected, while the _probe function of the PixCir is never called. This is the fragment from the dmesg kernel log:
...
<4>ekt2201_init
<4>ekt2201_i2c_ts_probe
<6>input: ekt2201 as /devices/i2c-0/0-0010/input/input0
<4>ekt2201:i2c_add_driver 0
<4>emxx_tsixcir_i2c_init
<4>emxx_tsixcir:i2c_add_driver 0
...
Note as the input device is created for the ekt2201 as the _probe is called, while the same doesn't happen for the pixcir.
Anyone with a clue?
thanks
Hi,
if anyone's interested in testing an experimental, pre-alpha version of the Open Source GB and Kernel build, leave a message here:
https://groups.google.com/forum/?hl=en&fromgroups#!topic/renesas-emev-osp/Mmd8r-YMaNo
Porting to JB has started: https://groups.google.com/forum/?fromgroups=#!topic/renesas-emev-osp/XSdLmPgD9Og
So I dived into the N7100 kernel, and tried a defconfig with Samsung's config. After a couple of setbacks in the form of toolchain related errors, I got it to create a zImage with the NDK/gcc4.4.3. But on booting, I noticed that the capacitative touch keys would no longer work. I'm guessing that I'm missing something in .config.
So where should I be looking in .config and the source tree for code related to the capacitative buttons?
I'm thinking we might work on a collaborative wiki for N7100 kernel source tree.
Samsung didn't share that with ANY of the "note II" variant source drops. However, you can pull the source directory from the i9300 kernel source. I forget the exact path (I'm at work atm), but it's something like drivers/input/keyboard/cypress
Take care
Gary
garyd9 said:
Samsung didn't share that with ANY of the "note II" variant source drops. However, you can pull the source directory from the i9300 kernel source. I forget the exact path (I'm at work atm), but it's something like drivers/input/keyboard/cypress
Take care
Gary
Click to expand...
Click to collapse
You grab it from the n8000 source which is full of updates for our devices. Or check my t889 kernel on github. I put the cypress drivers in a clean commit you can then create a patch from.
garyd9 said:
Samsung didn't share that with ANY of the "note II" variant source drops. However, you can pull the source directory from the i9300 kernel source. I forget the exact path (I'm at work atm), but it's something like drivers/input/keyboard/cypress
Take care
Gary
Click to expand...
Click to collapse
Except the HK source?
I was just looking into your git and found cypress in your initial commit and that you'd actually based the git on the HK source drop.
I untarred the HK source and found cypress there too. What did Samsung think? That international users could do without the capacitative buttons?
Edit: Compiled without a single error. Thank you.
Droidzone said:
I was just looking into your git and found cypress in your initial commit and that you'd actually based the git on the HK source drop.
Click to expand...
Click to collapse
You quoted me in that response... but I don't have any HK source drops in my github. The only Note II source drop I have is the SGH-i317 drop (and it's missing the cypress drivers as well.) Hmm..
garyd9 said:
You quoted me in that response... but I don't have any HK source drops in my github. The only Note II source drop I have is the SGH-i317 drop (and it's missing the cypress drivers as well.) Hmm..
Click to expand...
Click to collapse
lol..Sorry, I must've confused your repo with someone else's.
Anyway, check out the HK N7100 repo. It is complete and is about 10 MB more than the international stuff.
Droidzone said:
Except the HK source?
I was just looking into your git and found cypress in your initial commit and that you'd actually based the git on the HK source drop.
I untarred the HK source and found cypress there too. What did Samsung think? That international users could do without the capacitative buttons?
Edit: Compiled without a single error. Thank you.
Click to expand...
Click to collapse
Hi I'm having your problem, I'm download kernel source's and try bulding my First kernel for Gt N 7102, workinkg all except touchkeys.
Help me ?
Thanks bro
Sent from my GT-N7102 using Tapatalk 2
Did you use the HK source? What's the error?
Edit: Missed the N7102 part. See if cypress is present
Code:
cd /pathto/kernel
find . -iname cypress -type d
Droidzone said:
Did you use the HK source? What's the error?
Edit: Missed the N7102 part. See if cypress is present
Code:
cd /pathto/kernel
find . -iname cypress -type d
Click to expand...
Click to collapse
In original samsung source's cipress is not present but, I'm download from other source and insert in the correctly tree.
I'm modify kconfig file.
Try make menuconfig and cipress comparte to device input-->keyboard.
Make and install new kernel but touch keys not working.
Excuse my English and 'horrible.
Emblema said:
In original samsung source's cipress is not present but, I'm download from other source and insert in the correctly tree.
I'm modify kconfig file.
Try make menuconfig and cipress comparte to device input-->keyboard.
Make and install new kernel but touch keys not working.
Excuse my English and 'horrible.
Click to expand...
Click to collapse
Check your makefiles, and see that cypress has been asked to build.
I was looking at the mt_cpufreq.c file in the leaked kernel source for the MTK 6589 SoC (the one running at 1.2 GHz), at https://github.com/varunchitre15/MT...atek/platform/mt6589/kernel/core/mt_cpufreq.c
I noticed there were frequency settings above 1209 MHz defined there. The highest one, DVFS_F0_0 goes above 1.7 GHz.
The code that seems to define the maximum frequency starts on line 1573 of that file. It relies on the returned value of PTP_get_ptp_level to define it. Since PTP_get_ptp_level is just a external header on that file, I needed to search the git repo to find its definition. Turns out it is in this file mt_ptp.c. Pay special attention to line 1018 and below: https://github.com/varunchitre15/MT...ek/platform/mt6589/kernel/core/mt_ptp.c#L1018
There's this interesting bit:
Code:
#if defined(MTK_FORCE_CPU_89T)
return 3; // 1.5GHz
#else
By returning 3, this function tells the frequency scaling driver that the CPU can go up to a 1.5 GHz clock.
I think I read the 6589T also has a higher GPU clock, and it can handle 2 GB of RAM with no problems. I'm not saying that the difference between the 6589 and the 6589T is just a kernel config, in fact, the chip most likely got a new revision. This source code leaked before there was any talk of a turbo revision of the 6589, so I think it's safe to assume that on newer code trees there may be fixes for the 2 GB of RAM and a higher GPU clock.
But I wonder, if the 6589 (not T) is able to go up to 1.5 GHz or even 1.7 GHz. It would be good if someone could compile that kernel with some modifications to make it think it will be running on a 6589T (adding a #define MTK_FORCE_CPU_89T should be enough) and see if it works fine. Since that source code came from a Acer model, maybe it would be easier to start tests on that device (because we can be sure the source is 100% compatible). I think it may run, but it may heat a lot or it may drive the voltage of the CPU up to a point where the phone's PCB is not prepared to deliver, causing damage (maybe the T revision chip fixes this).
have you done any more digging or testing on this?
Just noticed that Jiayu have released a MTK6589T version of the G3S for only $5 extra - could it be just a simple software unlock?
thanks
My guess is that it really is only a kernel setting considering that the Jiayu G3S comes with both chips within 2 weeks. Besides this is a really easy way for mediatek to sell a wide range of SoC's (just varying the voltages/clockspeeds). The main thing I wonder if this also controls the clockspeed of the GPU, since that clock is different as well, considering the MKT6589 and MKT6589T.
http://www.ini3e.com/thread-2610-1-1.html
This chinese article confirms that the only difference is clock speed.
It would be really good if someone who can already compile custom kernels for any MT6589 phone, could make that little modification and check the results - worst case scenario, the chip would get too hot and break forever... but since it seems to have a lot of temperature monitoring in place, I don't think that would happen.
I could do it myself, but I'm not going to download 8 GB of Android source, on my limited home connection, just to apply the MT6589 patches available on GitHub, which probably wouldn't work out-of-the-box since they are for a Acer phone and mine's a Zopo 910.
Bothering people at Zopo to do it is out of question, since they are interested in selling the Zopo C2 which comes with this "turbo" kernel config.
As for the GPU clock, on this kernel source there doesn't seem to be anything controlling that based on the MTK_FORCE_CPU_89T define. But this may be just because this is an old source tree - the one they have been distributing for their clients more recently, for the 6589T, may already have the changes for the GPU clock.
The ideal situation would be to have a kernel which can allow all the possible CPU and GPU frequencies, and these would be controlled and limited on the userspace.
It would be awesome ! Please devs, do something for us
is there anyone working on this
kartikoli said:
is there anyone working on this
Click to expand...
Click to collapse
I looked at it shortly, and noticed a lot of stuff is missing to compile it.
But why not look at it yourself?
Start by git cloning the cm10.1 tree and compile for a device that is supported, to get an idea of how it works.
People who manage to compile the source, please post, because from there we can come close to building CM10.1.
The sources on that Github are meant to be merged with the JB source tree, you're aware of that when you say it's missing things, right?
Here are sources for another MTK 6589 device:
http://forum.xda-developers.com/showpost.php?p=43650399&postcount=10
gbl08ma said:
The sources on that Github are meant to be merged with the JB source tree, you're aware of that when you say it's missing things, right?
Here are sources for another MTK 6589 device:
http://forum.xda-developers.com/showpost.php?p=43650399&postcount=10
Click to expand...
Click to collapse
this group is working on a Cyanogenmod for mtk6589 devices...http://forum.xda-developers.com/showthread.php?p=41345830
http://wikogeek.com/ (Select CINK 5) complete kernel source (800 Mb compressed) with build scripts.
gowtham121 said:
http://wikogeek.com/ (Select CINK 5) complete kernel source (800 Mb compressed) with build scripts.
Click to expand...
Click to collapse
The Cink Five is a French MTK 6589 (not T) phone. Specs identical to my ZP910 except for the screen (higher resolution). Also, on Wiko Mobile's website they claim it supports USB OTG, which would be interesting to implement on other MTK phones.
Now downloading the sources (let's hope the connection doesn't fail, and that I don't spend all of the quota for this month... sigh).
EDIT:
I am now compiling this kernel tree, which is the most complete I've ever seen. Funny thing, this one doesn't have the 89T part I mentioned on the OP.
Ideas from http://forum.xda-developers.com/showpost.php?p=43861751&postcount=64 are proving to be useful when compiling this 6589 kernel (mainly the kalOidComplete errors thing).
Of course I'll never be able to get this kernel to boot on my Zopo, as I'd have to guess all of the kernel config including things like GPIOs... but getting a binary will at least be fun.
And in case you never looked into a Mediatek code tree, this thing is a horrible mess (something common to most OEMs and SoC manufacturers, anyway). And this is just the kernel... I don't even imagine how it would be if it also had the preloader, lk, etc. sources...
BTW, I just got a "**** Successfully built kernel ****" message, but then it says "Cannot open file /arch/arm/boot/zImage". Will have to investigate.
EDIT2: I managed to compile a kernel from the Cink Five source tree.
Interesting aspects:
This kernel source is more incomplete when it comes to CPU freq scaling than the one from the Acer Liquid E2 available on Github (and which is missing many of the scripts needed for compilation). For example, it doesn't know of any speeds above 1209 MHz. The source tree on GitHub is effectively more recent.
This kind of saddens me. It means that at some point, Mediatek wrote the code thinking the CPU wouldn't go over 1.2 GHz. Then they may have added the 89T-specific code later, only when they developed a physically different die that effectively allows for higher clocks. And so 6589 and 6589T may have more differences than just a kernel config.
You can see that new DVFS configurations were added to the source code at some point.
From mt_cpufreq.c @ Cink Five source tree (presumably before):
Code:
#define DVFS_F1 (1209000) // KHz
#define DVFS_F2 ( 988000) // KHz
#define DVFS_F3 ( 754000) // KHz
#define DVFS_F4 ( 497250) // KHz
#define DVFS_V1 (1200) // mV
#define DVFS_V2 (1150) // mV
#define DVFS_V3 (1050) // mV
#define DVFS_V4 ( 950) // mV
From mt_cpufreq.c @ Acer Liquid E2 source tree (presumably after):
Code:
#define DVFS_F0_0 (1703000) // KHz
#define DVFS_F0_1 (1599000) // KHz
#define DVFS_F0_2 (1508000) // KHz
#define DVFS_F0_3 (1404000) // KHz
#define DVFS_F0_4 (1300000) // KHz
#define DVFS_F1 (1209000) // KHz
#define DVFS_F2 ( 988000) // KHz
#define DVFS_F3 ( 754000) // KHz
#define DVFS_F4 ( 497250) // KHz
#define DVFS_V0 (1250) // mV
#define DVFS_V1 (1200) // mV
#define DVFS_V2 (1150) // mV
#define DVFS_V3 (1050) // mV
#define DVFS_V4 ( 950) // mV
This is just a very small part of the differences between the two code trees.
But even if the 6589 can't overclock, the 6589T apparently support a speed of 1.7 GHz (DVFS_F0_0)... unless they added that just for testing purposes.
All we now need is someone with a Cink Five to join this forum...Hopefully someone will be willing to test your kernel..
Anyone out there with a Cink Five???
Apparently the Cink Five is targeted at the French market, so probably one will have better luck asking in a French Android forum.
Any French speakers here?
Hi I'm a Wiko cink five user from France. I can give a try but I need stability. If you think it's stable enough, no problem, I can test it.
Thank you for your work !
frans0023 said:
Hi I'm a Wiko cink five user from France. I can give a try but I need stability. If you think it's stable enough, no problem, I can test it.
Thank you for your work !
Click to expand...
Click to collapse
To be honest, I think it wouldn't even boot, let alone be stable...
But if you can make a backup of the existing boot.img, then flash my one, and don't mind spending a hour or so without the phone, you could revert to the original boot.img if mine didn't work.
I don't know if Wiko released any updates for the Cink Five or if you have one installed, but that would make it easier to get the current boot.img for backup.
gbl08ma said:
To be honest, I think it wouldn't even boot, let alone be stable...
But if you can make a backup of the existing boot.img, then flash my one, and don't mind spending a hour or so without the phone, you could revert to the original boot.img if mine didn't work.
I don't know if Wiko released any updates for the Cink Five or if you have one installed, but that would make it easier to get the current boot.img for backup.
Click to expand...
Click to collapse
No problem for me, send me the file and I will give a try. Just tell me what I have to check, maybe you want the logcat ?
Only boot.img is written when you flash a kernel ?
If it's only that, I could boot on recovery (twrp) and backup the file with "adb pull", then push yours ?
I have to sleep, good night
Envoyé depuis mon CINK FIVE avec Tapatalk
Actually, boot.img is the name of the file on the computer, to be updated with SP Flash tools. I don't know if Wink ever released an update to be flashed this way (but since you have TWRP, probably you have used it).
On the phone (or with recovery), I know a kernel can be flashed with a update.zip.
Right now I don't even know how to create a boot.img file, I only have the zImage. I'm trying to get a way to repack my boot.img so that I can try it on my Zopo...
And then I'll have to solve the problem of the kernel modules (these may be harder to restore since they are in the system partition I think)...
Also:
Is it just me, or the Cink Five is the same as the Fly IQ451? Turns out the zImage I compiled was with the config for the s9201b board, which is the IQ451.
From the pictures of the case, it seems to be the same... but, with Chinese phones, the case means little.
The source tree came configured to use the tinno89_jb config but I changed it because I couldn't compile with it.
Aw yeah, you atleast tried to compile the kernel Let's hope the CINK Five guy can test your kernel (P.S: A engineering build please, so that I can access Thermal settings in MTK Tools)
gowtham121 said:
Aw yeah, you atleast tried to compile the kernel Let's hope the CINK Five guy can test your kernel (P.S: A engineering build please, so that I can access Thermal settings in MTK Tools)
Click to expand...
Click to collapse
I think you guys still haven't understood very well that my kernel most likely won't even get to the Android boot logo (I bet it will stop in the early kernel boot stages with some kernel panic).
Furthermore I need a boot.img for the Cink Five so I can put my zImage in there. Otherwise it will never be compatible.
Also, I think the build type is set on the Android building process. I can only build a kernel. I don't have a good enough internet connection to download the Android source, and my computer with 8 GB of RAM lacks the necessary 16 GB (!!!) for compiling.
I have studied the google's official documentation and I got the source with this command
$ git clone https://android.googlesource.com/kernel/msm
Now how can I get kernel source for specific Qualcomm Board like msm8994 or any other board
yuvraj_reckon said:
I have studied the google's official documentation and I got the source with this command
$ git clone https://android.googlesource.com/kernel/msm
Now how can I get kernel source for specific Qualcomm Board like msm8994 or any other board
Click to expand...
Click to collapse
You would need to look for the stock source used in the stock firmware from the manufacturer or carrier of the specific device in question.
I DO NOT PROVIDE HELP IN PM, KEEP IT IN THE THREADS WHERE EVERYONE CAN SHARE
git branch -a in the msm directory will give you the list of branches for particular boards / Kernel. you can checkout one of those branches using git checkout <branch name>
I want to compile android for a new device with that setup:
Code:
PLATFORM
OS Android 7.0 (Nougat)
Chipset Qualcomm MSM8940 Snapdragon 435
CPU Octa-core 1.4 GHz Cortex-A53
GPU Adreno 505
What compiler / env vars I should use?
Do I have emulator that I can config for that setup?
Thanks.
roman.mandeleil said:
I want to compile android for a new device with that setup:
Code:
PLATFORM
OSAndroid 7.0 (Nougat)
ChipsetQualcomm MSM8940 Snapdragon 435
CPUOcta-core 1.4 GHz Cortex-A53
GPUAdreno 505
What compiler / env vars I should use?
Do I have emulator that I can config for that setup?
Thanks.
Click to expand...
Click to collapse
https://source.android.com/source/initializing
DO NOT CONTACT ME VIA PM TO RECEIVE HELP, YOU WILL BE IGNORED. KEEP IT IN THE THREADS WHERE EVERYONE CAN SHARE
Droidriven said:
https://source.android.com/source/initializing
Click to expand...
Click to collapse
Thanks for the reply. I know that page but cant find there
the concrete differences in conifgs between different
chipset?
roman.mandeleil said:
Thanks for the reply. I know that page but cant find there
the concrete differences in conifgs between different
chipset?
Click to expand...
Click to collapse
What do you mean "configs between different chipset"? Setting up the environment is a universal thing, the differences should come when you sync repo to download your device's source code. Sync the correct source and you're ready to go, as far as I know.
DO NOT CONTACT ME VIA PM TO RECEIVE HELP, YOU WILL BE IGNORED. KEEP IT IN THE THREADS WHERE EVERYONE CAN SHARE