Related
I am trying to compile a clean AOSP ROM for my O2X, there is a few things I am having trouble with.
1. What is the significance of overriding the build prop with the following:
Code:
PRODUCT_BUILD_PROP_OVERRIDES += PRODUCT_NAME=lge_star BUILD_ID=FRG83G BUILD_DISPLAY_ID=GRJ22 BUILD_FINGERPRINT=lge/lge_star/p990/p990:2.2.2/FRG83G/lgp990-V10b.2ED2ADCFFC:user/release-keys PRIVATE_BUILD_DESC="star-user 2.2.2 FRG83G 2ED2ADCFFC release-keys"
Clearly its used for something, else CM7 wouldn't be doing it.
2. When I try to boot the compiled system (using either CM or EternityProject kernel), it simply stops at the LG logo, no logcat, no dmsg, nothing and after a while it reboots. What could I be missing?
I am using the init.rc from the CM7 p990 source, do I need some extra modules for it to boot? (I have added busybox and bash)
Tips on how I could debug the issue would be appreciated (any additional logs?).
I've managed to get froyo building and running from source (I used the nvidia branch and merged 2.2.2 myself)
I just used the stock LG kernel though
Got almost everything but gave up because it was a huge pain in the ass
You dont really need those
Are you trying to create the device tree yourself?
fallout0 said:
I've managed to get froyo building and running from source (I used the nvidia branch and merged 2.2.2 myself)
I just used the stock LG kernel though
Got almost everything but gave up because it was a huge pain in the ass
You dont really need those
Are you trying to create the device tree yourself?
Click to expand...
Click to collapse
Yea I am trying to create a device tree that will make AOSP compile and run. Its compiling just fine and it uses bits from the CyanogenMod device tree. The problem is that it wont run, it gives me no logcat output at all.
Here is the p990.mk:
http://pastebin.com/SxUZZDau
The boardconfig and proprietaries are identical to CM7. (Merged star-common and p990 configs since I have no use for p999)
Try using the stock kernel + init.rc - that is just use the boot image from modaco for now
You should be able to adb shell into it and check dmesg to see what's going on
fallout0 said:
Try using the stock kernel + init.rc - that is just use the boot image from modaco for now
You should be able to adb shell into it and check dmesg to see what's going on
Click to expand...
Click to collapse
Same problem, no dmsg or logcat.
Im gonna start from scratch now
About your first question (and sb correct me if I'm wrong), its just for market stuff to show up on the device, whilst making About>System display GRJ22 etc. etc.
L10nH34Rt said:
About your first question (and sb correct me if I'm wrong), its just for market stuff to show up on the device, whilst making About>System display GRJ22 etc. etc.
Click to expand...
Click to collapse
That would make it rather useless
KernelCrap said:
Same problem, no dmsg or logcat.
Im gonna start from scratch now
Click to expand...
Click to collapse
Also you HAVE to check out the code from nvidia because the tegra2 has a bug in the CPU and their code base has a work around for it if you don't the dynamic linker doesn't work and you'll find every executable will segfault ( when you flash your system image from recovery try and go into system/bin and run. /ls for example)
In the board configuration you'll notice a HAVE_TRGRA something or rather line
check out bionic that's where that fix is applied
fallout0 said:
Also you HAVE to check out the code from nvidia because the tegra2 has a bug in the CPU and their code base has a work around for it if you don't the dynamic linker doesn't work and you'll find every executable will segfault ( when you flash your system image from recovery try and go into system/bin and run. /ls for example)
In the board configuration you'll notice a HAVE_TRGRA something or rather line
check out bionic that's where that fix is applied
Click to expand...
Click to collapse
Im gonna try to build using the 2.3.3 gingerbread-tegra source from nvidia
EDIT: Same problem, when running something i get "Segmentation fault". I am using the gingerbread-tegra source from nvidia :S
I have the following in BoardConfig.mk:
TARGET_HAVE_TEGRA_ERRATA_657451 := true
Busybox is working fine though :/
Shouldn't the source from nvidia allow me to execute the compiled binaries without giving segmentation fault?
The funny thing is that busybox is working, and I can access dmesg through busybox.
KernelCrap said:
That would make it rather useless
Click to expand...
Click to collapse
It wouldn't if you couldn't access the market now would it?
L10nH34Rt said:
It wouldn't if you couldn't access the market now would it?
Click to expand...
Click to collapse
I'd rather have it boot atm.
Still same problem even though I have moved to the gingerbread-tegra source
IMHO its a wasted effort. You'll have trouble getting camera/hdmi etc.etc to work... Just clone CM repo and cherry-pick from there.
L10nH34Rt said:
IMHO its a wasted effort. You'll have trouble getting camera/hdmi etc.etc to work... Just clone CM repo and cherry-pick from there.
Click to expand...
Click to collapse
For starters I don't care about camera or hdmi, I want a clean AOSP rom - CM is great, but it includes too many extra features for my taste.
Are you sure you checked out the tegra branch from nvidia? gingerbread-tegra?
I could only get froyo to boot myself - gingerbread would crash inside surfaceflinger with the froyo graphics libraries. I think the cygonenmod guys had access to leaked gingerbread libraries.
Hi guys, where i can find sources from nvidia ? I only found the kernel git...
Thank's
fallout0 said:
Are you sure you checked out the tegra branch from nvidia? gingerbread-tegra?
I could only get froyo to boot myself - gingerbread would crash inside surfaceflinger with the froyo graphics libraries. I think the cygonenmod guys had access to leaked gingerbread libraries.
Click to expand...
Click to collapse
I am sure and I am using the leaked libraries too. Anything else i can try?
Sent from my Optimus 2X using XDA App
If all the executables are seg faulting then you won't even get to logcat and the shell won't run unfortunately.
Try this...
find all the blocks of code in the source that have the #ifdef HAVE_TEGRA..ERRATA
and remove them
That way you'll definitely know that its compiling that section of the code. That has to work
fallout0 said:
If all the executables are seg faulting then you won't even get to logcat and the shell won't run unfortunately.
Try this...
find all the blocks of code in the source that have the #ifdef HAVE_TEGRA..ERRATA
and remove them
That way you'll definitely know that its compiling that section of the code. That has to work
Click to expand...
Click to collapse
Will try that once I get home later today
I commented it out the following places:
Code:
./bionic/libc/Android.mk:497: ifeq ($(TARGET_HAVE_TEGRA_ERRATA_657451),true)
./bionic/linker/Android.mk:46:ifeq ($(TARGET_HAVE_TEGRA_ERRATA_657451),true)
Still not booting, now the error is (from logcat):
Code:
exec '/system/bin/sh' failed: No such file or directory (2)
But /system/bin/sh does exists, with the following permissions (also tried with chmod 777):
Code:
-rwxr-xr-x 1 root shell 82840 sh
If I do adb shell when in recovery and executes the binary files, they all seg fault (except for busybox) :/
Here is my BoardConfig.mk:
http://pastebin.com/ZsmJp0v1
Status: working!
The tar file has some binaries which work.
If you aren't a developer you probably don't care about this.
I'm working on getting FUSE working on my P500, and thought I'd share my experiences. FUSE is a mechanism to mount lots of filesystems (like fusecompress, sshfs, or unionfs) without needing to recompile the kernel or patch the kernel. My need is to have a writeable /system that can be 'cleaned' back to the original state with the push of a button.
The following details my initial attempt. I've had better success as detailed here:
http://forum.xda-developers.com/showpost.php?p=14619814&postcount=3
Steps:
1) get the kernel source and build it with the FUSE module enabled. this provides fuse.ko which should be usable on virtually all 2.6.32.xx kernels for the P500
2) get the fuse userspace libraries and compile. I had to use a very old version (2.4.2) because newer versions need support for statvfs which bionic doesn't have. I had to patch in apthread_rwlock which isn't in bionic for froyo, and a few other cleanups.
3) get a fuse filesystem to test. I'm using unionfs-fuse. I had to use a very old version (0.11) to go with my very old libfuse from (2)
4) Try it out:
* move the files from 1-fuse.ko, 2-fusercompress, and 3-unionfs to the p500
* become root
* make sure all files are executable and in the PATH
* load the kernel module: insmod fuse.ko
* try to use fuse: ./unionfs --roots=/sdcard /mnt/asec
The last step seems to work, but ps shows no unionfs process.
Next I try with debug enabled:
* ./unionfs --roots=/sdcard /mnt/asec -d
This gives a segfault
Lastly I try with strace to find the segfault:
* strace -ff -F ./unionfs --roots=/sdcard /mnt/asec -d
now the code no longer segfaults, and the mount works.
So my goal now is to figure out why it segfaults when not run via strace.
I found someone who got this working on the desire:
http://forum.xda-developers.com/showthread.php?t=763908
(I tried their binaries just for fun, but they're not compatible with the P500)
They did it by using scratchbox which imports glibc. I was trying to do a native bionic build, but it may be easier to try scratchbox, as it would let me have a modern libfuse. If I can't figure out this segfault, then I'll go there next. Has anyone tried building for the P500 using scratchbox?
Sweet idea man!
Ok, I have fuse fully working using the latest fuse version 2.8.5
In order to build, I got a copy of statvfs.c and statvfs.h from the shieldroid project. I had to modify the definition of f_fsid to be 'unsigned long' and adjust statvfs.c appropriately I also ended up disabling multi-threaded support in fuse as it made it made it easier to compile.
With a bunch of monkeying around, I got libfuse compiled with the apropriate utils. I verified the example filesystems worked, then went ahead and built unionfs-fuse (latest version).
Using the tar file in the 1st post you can now:
* copy all files to /data and make executable
* add the relevant directory to your path
* insmod fuse.ko
* ./unionfs /sdcard:/system/bin /data/local/tmp
* ls /data/local/tmp (should contain the contents of /system/bin and /sdcard in one merged dir)
When I get a chance, I'll build an installable zip that puts the files in the right place for easy use.
FYI, my goal is to create a utility that can temporarily enable root on a phone and then disable it when done. My idea is to move the 'su' command out of /system/bin onto /data, and then use unionfs to teporarily add it back onto /system/bin on command. When done, you'll be able to click to unmount it again, bringing the phone back to an unrooted state.
It'll work sort of like the 'temporary root' capability of z4root, but will be undoable without a reboot (it also won't inculde the code to actually break out of a stock ROM, the goal is to be able to toggle root on an already rooted phone).
I need this capability because I run some enterprise software that doesn't work on a rooted phone, and I don't want to be constantly booting to recovery for root tasks.
i dont understand most of wat u saying, but good going man.
im also computer science student but hate studying =(
Ok, this goes out to any and all DEVs out there. We all know by now that we (some people not me) can run linux within android using the loop mounts, vnc viewer, etc... Now how about REPLACING android with a linux distro like debian or maybe even what these guys use http://openpandora.org/ ..... its linux based and has an arm CPU.. Any way I'm talking about flashing over android wiping the internal sorage and installing linux on it or even using the boot loader to flash over android???, of course this is getting rid of any 3g connection and phone usage. I understand that, I am talking about giving our RETIRED droid 1s a use. I wanna see my D1 run a linux distro in full hardware mode NO MORE ANDROID. now dont get me wrong I love android as much as the next guy, but why now flash something else to it???
Another link I found, this is for windows mobile but hey, similar idea.
http://wing-linux.sourceforge.net/trac/wiki/FAQ
P.S. I AM NOT A DEV JUST AN IDIOT WITH AN IDEA!! my ignorance is NOT bliss and I would love to know if this is even possible. I'm willing to help ANY WAY I CAN. lol
THANKS!!!
I don't think it's possible. Because of the locked bootloader we're forced to use Motorola's bug-ridden Kernel and as Android's Kernel is a heavily modified Linux Kernel it most likely won't run an ordinary Linux distro.
But I'm no expert either
Milestone is locked, Droid is not ...
Maybe this will help (if you don't know german, you can use Google Translate):
http://www.android-hilfe.de/anleitu...debian-chroot-mit-lxde-auf-dem-milestone.html
Thanks for the replies. Again the USA Droid 1 is nit locked or at least has been completely unlokcked. I've read a million "run debian on android" posts... I don't need to know German to see that the post above is the same thing. Like I said before I have NO INTEREST IN RUNNING LINUX ON ANDROID. I want TO REPLACE android completely with Linux to make a device like the openpandora handheld from my first link.
Any one up to the challenge????
Thanks again!
Really, no one??
I've been running Debian on my Droid booting from the SD card for a while, more recently trying Arch Linux after realizing that anything optimized for ARMv7, VFPv3, NEON, or Thumb-2 (I'm not sure which unfortunately) won't boot because of a page fault or something. Ubuntu, MeeGo, and Angstrom just kernel panic and don't give any useful information even at the highest debugging level. I'm back to using an ARMv5te Arch Linux build (http://archlinuxarm.org) although I could just as well use Debian. I really wish Ubuntu worked for multitouch.
Well, I got the touchscreen calibrated! I forgot that my screen rotation hack only rotated the framebuffer It's stuck in portrait for now. I decided to use mtev (MeeGo's multitouch X11 input driver) after being fed up with evdev's aversion to being rotated, but now that I'm back to the stock portrait rotation evdev should work fine.
If you want instructions, either PM me or wait until I post a full guide and/or my patched kernel tree. It's not super difficult, but it's a lot of command line use and compiling.
What works:
*CPU and SD card (obviously )
*Touchscreen (single touch/portrait only for now)
What needs work:
*Keyboard mapping is wrong, the number keys and symbols don't work. I need to figure out how Android handles Alt. I had to patch the GPIO keyboard driver because the keyboard worked in the console but not in X11, which expects EV_SYN.
*Sound is OSS only, but there is no mixer device, so aumix is useless and there's no volume control or sound output.
*I haven't tried the SGX driver, so I can't comment on hardware acceleration yet.
*Battery charging relies on battd, which is a proprietary Android binary from Motorola. It might run on Debian with "ls -s / /system" and the creation of the socket it expects (init.sholes.rc I believe), but I'm not getting my hopes up.
Untested:
*WiFi (needs firmware, but should work)
*Bluetooth
*Calls/data
*Sensors (although the accelerometer and compass seem to be recognized by evdev)
I'll upload my kernel, either as patches or on GitHub (or binaries if there's enough demand) once I get around to fixing the keyboard issue. Fingers crossed that I don't get carried away cleaning up the kernel, or worse yet, give up...
This is incredible news! I'm so glad to hear it. Of course you are running in hardware mode? Not through android? If so I'm super exited about this! I love arch Linux, I run it on all my computers and I understand it more than any other distro I've tried. I would love to give it a shot, though I'm not that great at compiling I think it will be worth it.
As I said before I am in no way a dev or even a Linux pro, but I can get by and if theres any way I can help this project become "complete" I'm willing to learn what ever is necessary to do so.
Also were u able to get an x server running? Possibly a DE? That would be crazy awesome.
Thanks a million, looking foreword to work with you and make this happen!
For anyone who's interested, I put together a quick package containing my kernel patches, configuration notes, and an automated kernel build script.
It's pretty self-explanatory aside from the installation of the root filesystem, which I leave up to your imagination
As I've probably repeated several times now, I've run Debian and Arch Linux natively on my Droid with my patched kernel (no chroot or VNC "hack"). With enough determination, it's possible to run practically anything on it - personally, I'm determined to get Ubuntu on it.
WOOT... insta-fail for me LOL you have it set up to work in specific directories? I thought i placed everything right but I guess not.. I got an error on the first line of output lol
Welcome to gTan64\'s lazy Droid kernel compiler\!
build.sh: line 5: [: missing `]'
build.sh: line 5: -z: command not found
build.sh: line 5: -z: command not found
Entering $WORK
Applying patches
build.sh: line 16: /*.patch: No such file or directory
mv: cannot stat `/arch/arm/boot/compressed': No such file or directory
cp: cannot stat `/compressed': No such file or directory
Setting up build
cp: cannot stat `/GNUmakefile': No such file or directory
cp: cannot stat `/sholes_config': No such file or directory
FIXME! No numeral or symbol input until someone finds a third-level
modifier key to use instead of AltGr for the keyboard.
cp $MISC/defkeymap.map $K_SRC/drivers/char
Building kernel!
make: *** No targets specified and no makefile found. Stop.
Kernel build done!
You have done something amazing, I'm just to stupid to figure it out my self. XD
I can understand why you would want ubuntu with all its pre-setup glory and support for nearly all hardware but dont you think its a little heavy for this hardware? lol I doubt you would try to use the Unity UI but I doubt that alone would even fit in ram LOL you could just use the ubuntu packages and drivers with debian maybe?
THANKS AGAIN!!!!
That's what happens when I don't test my own scripts
I was thinking too much about making it easy. Just try applying my patches to your tree manually.
Code:
cd kernel_src
patch -p1 <../patches/first.diff
patch -p1 <../patches/next.diff #and so on...
#Obviously, substitute the paths of the actual patches ;)
#Finally, copy sholes_config to .config and invoke 'make'.
It's actually much simpler than my script makes it, and you don't need my convoluted folder setup.
UM lol I cant find any patch files? are they in the ZIP? or do I have to obtain them elsewhere?
I can tell this is gonna be a hell of a project LOL
THANKS!!!
"Duh" moment - I used the .diff file extension instead of .patch
I'm such a scatterbrain... Remind me not to release anything before testing
EDIT: Strike that, reverse it. They're .patch files.
I AM BLIND lol I found them... BUT.. when I try to patch. terminal just seems to lock up... I press enter, the curser moves to the next line but it doesnt do anything. lol how long should 1 patch take?? its been about 10 minutes and I leave it be for now
EDIT:
LOL i forgot the "<" in the command OOPS but I'm not seeing any .config file or folder in the source to copy the sholes_config too. yes I have show hidden files on.. PS unfortatly usinf Fedora 15 gnome 2...... (on server so I can work on this anywhere through vnc )
THANKS
When I said "copy sholes_config to .config", that's literally what I meant. .config is the Linux kernel build configuration file, not a folder.
'cp /path/to/sholes_config .config' from your kernel tree should do it.
Oh, I forgot to mention that you need an ARM cross-compiler/toolchain. I use the gcc-arm-linux-gnueabi package in Ubuntu, but I don't know what the Arch equivalent is off the top of my head.
The "GNUmakefile" is just a convenience, it exports "ARCH=arm CROSS_COMPILE=arm-linux-gnueabi-" before calling the actual makefile. If you don't have an ARM cross compiler in your path, it won't build at all. If you do, modify CROSS_COMPILE to point to it - on some systems it's called arm-unknown-linux-gnueabi, etc. etc.
Let me know if you get anywhere before it errors out.
OK I put the config file in the kernel_src but unfortainatly I'm using Fedora 15 XP I installed arm-gp2x-linux-gcc which is the fedora arm gcc I guess... lol so wha would I edit in the .config to make it point to arm-gp2x-linux-gcc. Its not my first attempt to compile an android kernel but I've never been succesfull LOLOL
thanks
I'm pretty sure that toolchain won't work - the GP2X was ARMv5 IIRC, so unless it's a newer build of GCC (4.3 or higher) and you're absolutely sure it supports the ARM EABI and the ability to generate ARMv7 instructions (which is unlikely if it's pre-4.1), I would recommend a newer toolchain.
CodeSourcery makes a pretty solid one: https://sourcery.mentor.com/sgpp/lite/arm/portal/release1803
If you end up using it, you can change the second line in GNUmakefile to "CROSS_COMPILE=arm-none-linux-gnueabi-". Don't worry about changing .config.
Well I DLed the linux/GNU installer and it gave me "arm-2011.03-41-arm-none-linux-gnueabi.bin" LOL
so I was thinking would it just be easier to just use my lappy with crunchbang (debian based) so I can follow ur instructions with deb / apt-get LOL fedora is a ***** and it dont have what I need... I cant find an arm v7 cross compiler for it
thanks
.bin files are the Linux equivalent of Windows EXE installers. Run it like this:
Code:
#cd /path/to/installer.bin
./installer.bin
#substitute the actual name, of course.
It needs root permissions if you want to install it to /usr/local (sudo ./installer.bin or su -c 'sh installer.bin'). You can also install it to /usr, but that will make it more tedious to uninstall later. If you decide to install it in your home folder - say ~/toolchain - you don't need root permissions, but you will need to add ~/toolchain/bin to your path:
Code:
export PATH=$PATH:$HOME/toolchain/bin
To be honest, I prefer Ubuntu to Fedora/OpenSUSE/$RPM_distro, so if you have access to a Debian or Ubuntu system, using that would make my life easier
Omg duh... the "non_Linux" part of the name true me off lol i'll start doing this on my laptop it's deb based crunchbang. I prefer anything over rpm distros too but it was a quick painless install on my server XP thanks ill try this when I get home
Sent from my DROIDX using XDA App
Planning this for sometimes. So much work ahead. Just studying the new code cause me a lot of time. I think I have to rush it out so that someone may share my burden. Don't have much time to polish it. I'm trying to provide all commands, files and links. Please correct me if something is wrong or missing. Hoping this could be a stepping stone for those interested. If you like challenge, it'll be fun!
1. Introduction
If you have watched "Top Gear", you would have heard Jeremy Clarkson said "How hard could it be?". Why is it so hard to build a rom for our tab?
When I first play with rom building, it was quite straightforward. I recuilt my retired PC (Celeron 1.2G with 3G RAM, 40G HD) and follow the guild for building PA. Installing Ubuntu 12.04 and then the necessary packages. Everything is ready in a few hours. It was really challenging your patience to repo sync the source. After a few retries, I finally completed the download in two days! After two hours, my harddisk is full! Then I collected another retired 80G HD from a friend. Re-install everything and build again. Building requried 7 hours on my antique PC. With only a few minor errors to correct, my first rom pa2.55 was out. So how hard could it be? It was our predecessors ( @ohanar, @locerra, etc.) :good: who had already done lots of hardwork on device trees and kernel! If Samsung has really released the hardware related sources, there will not be so much troubles for us. At the time Android 4.2 launch, we have a new problem. Our tab's CSR chip doesn't support Bluedroid. See the Nexus family, everything is ready. Rom building will never be so pain in the ass as us.
1.1 What's wrong with our hardware
GT-7.7 and Plus use the exynos4 smdk4210 board with Mali-400 GPU. Quite a number of devices have similar hardware. The main problem we have would be:
1.1.1 CSR bluetooth chip.
CSR chip doesn't support bluedroid which become standard after Android 4.2. Bluez.org has provide a solution for 4.2 early this year but porting to 4.3 and 4.4 may take some time.Working on it now. Anyone has experience on bluetooth are welcome to help.. Bluez working now. See my Thread How to port Bluez to Android 4.4. Cheers!:highfive:
1.1.2 Athero wifi chip.
Samusng's wifi driver has issues on 5G support and SOD. I'm using the modified kernel driver from Samsung's 4.1.2 source. Athero provide drivers for 3.1 to 3.10 linux kernels but they have no wakelock support. I also wrote a drive base on Athero's source and add wakelock support but have no time to polish it now. Athero's chip doesn't have firmware for AP as Boardcom. Which is also the way Android platform providing wifi tethering in framework. To support wifi tethering we have to change the framework and the driver. Anyone has experience in linux HostAP are welcome to help. Wifi tethering is ok now! Tutorial will be in Post#2 soon.
1.1.3 modem chip (xm6260).
RIL stand for (radio interface library). Most of the phone manufacturer would provide their own vendor lib. The only working blob now is Samsung's ICS propietry libsec-ril.so which is obselete now. The one from stock 4.1.2 not working but no time to trace. It also don't seems to have much difference. This library handle the AT commands issuing to the (baseband to) modem chip. xm6260 has a complete set of command manual. It should not be too hard to rewrite if we can hack the ports. However it is not worth to splend too much time on this. Those has experience in AT commands are welcome to help.
RIL details: http://www.kandroid.org/online-pdk/guide/telephony.html
1.2 Where to get the source?
1.2.1 Samsung's source for our tab
The best source would be from Samsung http://opensource.samsung.com/. Sadly, the source from Samsung only cover the kernel. Our tab only have source up to 4.1.2 and kernel source 3.0.31 from Samsung.
1.2.2 Samsung's source for other devices
Luckly, we have similar architecture to some newer Samsung models. Their source is our major source now. eg. We can use the Mali driver from Note8 and/or i9300.
1.2.3 Manufacturere's web site.
The other source would be from the chip manufacturer. eg. Athero has source for linux.
1.3 What is going on at boot?
1.3.1 Power ON
a) When power button is long pressed (should be released after the logo shows otherwise it will reboot again every 6 seconds). The manufacturer's bootloader would show the vendor's logo. At that moment, bootloader would sense if there are key being pressed. Standard Android will go to boot menu when pwr + vol- is pressed. Samsung would go to download mode (something like fastboot) and accept pwr + vol+ to start recovery mode. If no other key is pressed, it would be a normal boot.
b) When USB is plug in (PC usb port or charger) at power off, the tab will be turn on silently. It is in the LPM (low power mode). According to lpm.rc (executed by Init, will be explained in 1.3.3), CPU will be set to POWERSAVING and the program charger is running. It should only handle simple tasks like Detecting the pwr button to show the charging images. If the pwr is long press when showing the charging image, it will reboot as in a).
3) Lollipop LPM: charger is link to the program /sbin/healthd. The process is similar to b) but lpm.rc is merged to init.rc.
1.3.2 Boot image
For our tab, there are two parts for a boot.img or a recovery.img. kernel binary and initramfs. As the name imply, Init-Ram-FS is the initial file system loaded into ram when the devices boot. Kernel binary provide the hardware drivers libraies which communicate with the hardware and baseband. At normal boot or recovery mode, the image (in /boot or /recovery partition) will be decompressed and loaded to memory. Initramfs will be loaded in root(/). It would consume a few hundreds MB of RAM and that's why our free ram is only 7xxM.
In LPM, it would depends on the last boot state. If it was powered off from recovery, recovery.img will be loaded and handle the LPM. If it was powered off from platform, boot.img will be loaded and handle LPM. That's why It might show different charging images if the charging images for reovery.img and boot.img are different.
Stock cook roms can unpack and modified the initramfs from the existing zImage. It is created in system/core if we build it from source. Here is a tool which can unpack/repack our boot.img or recovery.img.
1.3.3 Init
There is the Init program in every initramfs. This is the first program to run according to the sequence in the init.rc (or lpm.rc when in LPM). Partitions will be mounted and symlinked, default values and permissions are assigned, services will be started, etc. The source of Init is in system/core. The last thing Init do would be the continuous eventloop to process events.
2. Getting Start
To start building, we need the building environment. Many tutorials have alreay covered the details so I just list my suggestions. The guide from Google is a bit outdated but really helpful. http://source.android.com/source/building.html
Basic knowledge of linux is required. All commands are run in a terminal.
2.1 Hardware
A decent PC with as much ram as possible. Thanks for the donations, I can upgrade to a better i5 notebook with 4G ram and a 500G partition only for building. It would take 3+hrs. for a clean build. RAM and Harddisk speed are essential. A desktop PC with same configuration will be faster since notebook harddisk is slower. I plan to add more RAM when the price drop (seems to be a long wait )
RAM: 4G or above is recommended.
Harddisk space: 30G+ for system, 60G+ per rom, size of ccache (50G-100G), Roughly, a partition with at least 120G for one rom. Bigger is better.
A decent internet connection is essential. It will take days to sync the source if you only have bandwidth of 2M or less .
2.2 OS
64bit Ubuntu Desktop 12.04 or above. Recommend 13.04 or above. I'm using the latest 14.04.
2.3 Install the required packages
These packages are required during building and packing. (eg. gzip is required to compress the kernel binary)
type command: (have a beer or coffee. It will take a while)
Code:
[I][COLOR="Green"]For Ubuntu 12.x and 13.x:[/COLOR][/I]
sudo apt-get install git-core gnupg flex bison gperf build-essential zip curl zlib1g-dev zlib1g-dev:i386 \
libc6-dev lib32ncurses5-dev ia32-libs x11proto-core-dev libx11-dev:i386 libreadline6-dev:i386 lib32z-dev \
libgl1-mesa-glx:i386 libgl1-mesa-dev g++-multilib mingw32 tofrodos python-markdown libxml2-utils xsltproc \
readline-common libreadline6-dev libreadline6 bzip2 libbz2-dev libbz2-1.0 libncurses5-dev lib32readline5 \
lib32readline-gplv2-dev lib32readline6 libreadline-dev libreadline6-dev:i386 libreadline6:i386 pngcrush \
libghc-bzlib-dev lib32bz2-dev libsdl1.2-dev libesd0-dev squashfs-tools schedtool libwxgtk2.8-dev python
sudo ln -s /usr/lib/i386-linux-gnu/mesa/libGL.so.1 /usr/lib/i386-linux-gnu/libGL.so
[I][COLOR="Green"]For Ubuntu 14.x and 15.x:[/COLOR][/I]
sudo apt-get install git-core gnupg flex bison gperf libsdl1.2-dev libesd0-dev libwxgtk2.8-dev squashfs-tools \
build-essential zip curl libncurses5-dev zlib1g-dev pngcrush schedtool libxml2 libxml2-utils xsltproc lzop \
libc6-dev schedtool g++-multilib lib32z1-dev lib32ncurses5-dev lib32readline-gplv2-dev gcc-multilib
2.4 Insatll android-sdk
Download (for Linux 64-bit of course): http://developer.android.com/sdk/index.html
Extract the file to a directory (eg. ~/android-sdk)
Add the path to ~/.bashrc
type command:
Code:
sudo gedit ~/.bashrc
at the bottom add the path
Code:
export PATH=~/bin:~/android-sdk/tools:~/android-sdk/platform-tools:$PATH
Quit and restart terminal to commit the changes.
Verify the path and update the SDK
type command:
Code:
android
2.5 Install java
If you have installed eclipse, openJAVA is ready. Some say it is not suitable for android development.
I've tried both and I found no appearing difference.
To install SUN JAVA (Ubuntu 13.04 or above)
type command:
Code:
sudo add-apt-repository ppa:webupd8team/java
sudo apt-get update
sudo apt-get install oracle-java6-installer
For Lollipop: openjdk 7 is required
Code:
sudo apt-get install openjdk-7-jdk
2.6 Install repo
type command:
Code:
mkdir -p ~/bin
curl http://commondatastorage.googleapis.com/git-repo-downloads/repo > ~/bin/repo
chmod a+x ~/bin/repo
2.7 Add usb dev rules
Which are required for adb and fastboot, eg. View attachment 51-android.zip
type command:
Code:
sudo cp 51-android.rules /etc/udev/rules.d
sudo chmod 0644 /etc/udev/rules.d/51-android.rules
2.8 Enable ccache
type command:
Code:
sudo gedit ~/.bashrc
add lines for CCACHE
Code:
export USE_CCACHE=1
export CCACHE_DIR=~/android/.ccache
Quit and launch the terminal again to commit the changes
ccache will cache the object files of gcc for next build, it would increase the speed of subsequent builds alot.
After source is sync in 2.11 and before build, we can set the size of ccache first.
type command:
Code:
~/android/rom/prebuilts/misc/linux-x86/ccache -M<size>
2.9 Init the source
Make a new working driectory for your rom
type command: (eg. ~/android/rom)
Code:
mkdir -p ~/android/rom
Each rom has its own manifest.xml listing all the required projects.
It is usually in the project android (manifest for PA). The readme of this project would provide the command to init.
eg. CM11.0: https://github.com/CyanogenMod/android/tree/cm-11.0
type command:
Code:
repo init -u git://github.com/CyanogenMod/android.git -b cm-11.0
It will fetch the manifest.xml and put it into a hidden directory .repo in the current directory (~/android/rom/.repo)
2.10 preparing local_manifest.xml
It is not a good pratice to change the manifest.xml directly. If the ROM has change its manifest, you would require to init again. The right way is putting everything your device needed in a seperate file .repo/local_manifests/local_manifest.xml. The device trees and kernel source for GalaxyTab 7.7 and Plus : https://github.com/danielhk?tab=repositories
eg. cm11.0 for p6800
Code:
<?xml version="1.0" encoding="UTF-8"?>
<manifest>
<project name="danielhk/android_hardware_atheros_wlan" path="hardware/atheros/wlan" />
<project name="danielhk/android_device_samsung_p6800" path="device/samsung/p6800" />
<project name="danielhk/android_device_samsung_smdk4210-tab" path="device/samsung/smdk4210-tab" />
<project name="danielhk/android_kernel_samsung_smdk4210" path="kernel/samsung/smdk4210" />
<project name="danielhk/proprietary_vendor_samsung_smdk4210" path="vendor/samsung" />
<project name="CyanogenMod/android_hardware_samsung" path="hardware/samsung" />
</manifest>
Note: Settings will follow manifest.xml. cm-11.0's manifest has default remote="github" and revision="refs/heads/cm-11.0" so we can omit the same entries here. For other roms, see the defined names in manifest.xml and change accordingly. You can also add other devices trees here if you want to build them for the same rom.
2.11 sync and build
To fetch the source (specified in manifest.xml and local_manifest.xml) to working directory (~/android/rom)
type command:
Code:
repo sync -j32
-J specify the no. of threads to download. I have a few trials. Seems 32 works for me.
When it is done. You will find that ~/android/rom is not empty now.
You may set the ccache size now:
type command: (eg. 50G)
Code:
~/android/rom/prebuilts/misc/linux-x86/ccache -M50G
Only if you want to change size, otherwise it only requires once.
To build a rom, change to the rom directory. Usual commands: (eg. for CM roms)
Code:
cd ~/android/rom
. build/envsetup.sh
lunch cm_p6800-userdebug
mka bacon
CM need to download a prebuilt (Termianl.apk). To simplify this routines, I modified the build script from PA to cm View attachment rom-build.zip.
Copy it to the rom directory, type command:
Code:
cd ~/android/rom
./rom-build.sh p6800
After building is complete with no error, the rom will be output to ~/android/rom/out/target/product/p6800.
3. What can we do? :laugh:
3.1 Kernel
Since Samsung provide the complete kernel source, working on kernel would be comparatively speaking much easier.
When we say kernel, we mean the boot.img or zImage which already include the initramfs. This image will be flash to the /kernel (/dev/block/mmcblk0p5)
Hardware and flags are defined in defconfig inside <kernel dir>/arch/arm/config. It is the Linux standard.
3.1.1 difference between stock kernel and the kernel in custom rom
The main difference between the two kernels is the location of kernel library modules. Libray modules are mainly for network and storage drivers. By building them as modules, they can be stoped and restarted in case of critical errors.
Stock kernel put it in /lib which is part of the initramfs. We need to do extra steps to manually copy them to the initramfs after the first build and rebuild again. If you want to build the stock kernel, there is a very good tutorial.
Custom rom put the kernel library modules in /system/lib. Thus boot.img need to match those modules, otherwise they will not work properly.
For custom roms, which defconfig to be used is defined in device tree.
eg. For p6800, in the file device/samsung/p6800/BoardConfig.mk. There is the line:
Code:
TARGET_KERNEL_CONFIG := cyanogenmod_p6800_defconfig
It specify which defconfig file for the kernel
I wrote a View attachment build-k.zip scirpt to simplify the building of kernel alone. Copy it to the rom working directory (~/android/rom)
eg. For p6800, type command:
Code:
./build-k p6800
New boot.img will be output to the same directory as rom if no error.
3.1.2 Add/modify O/C, U/V, GOVERNORS, hardware drivers, etc.
As mentioned in 1.2, we can copy/modified the source from i9100 and n7000 kernels. For example, Dorimanx and Slyahkernel provide many useful code for us. The source of i9300 from Samsung may also provide some drivers update (eg. Mali)
Ref: Exynos4210 cpu menu., Mirror download
3.1.3 Default Orientation
Why the incall screen always landscape?
Many apks assume the default orientation is portrait but our tab's default orientation is landscape. It is a kernel behaviour so that recoveries (including the stock 3e recovery) are all in landscape.
To avoid this odd behaviour of apks like inCall screen, changing the default orientation in kernel is the ultimate solution.
It is already in my do list for a long time but trying to keep up with the android plaftform already occupy all my spare time...
I already have some idea how to do. Anyone interested to help on this can PM me. Already fixed in the platform
3.1.4 Update the kernel source
The existing kernel is base on Linux 3.0.x. It is outdated but I don't think Samsung would bother to update it. I had tried starting with the 3.4 source from Google but don't have enough time to carry on. I have filled some drivers already. Will take another shot later if I have time...
Anyone have experience in Linux kernel are welcome to help. No more guides are there, this would be a great challenge but fun!
3.2 device tree
Every device requires a device tree which:
a) Specify the hardware
b) define some overrided library. (eg. sensor, camera)
c) provide the extra files to be put in the initramfs. (init.<device>.rc, fstab.<device>, etc.)
If there are a few models variants, common parts would be put in a seperate common device tree (eg. smdk4210-tab for our tab). In this way, changing can be made once for all variants.
Most of the work on porting for a specific ROM would be on the device tree. Same as the kernel, we can get some update by modifying from other devices. CM's device tree for i9100, n7000 and i9300, etc. With a good device tree, we can port to any ROM.
3.3 Bluetooth support after 4.2.
In 4.2.2, I changed the source of framework, system_core, phone apk and device tree to work with bluez. It was a huge job! Hundreds of lines had been changed. Many files are replaced. It was quite clumsy but was the only way then. This way may not work for 4.3 onward.
Here are the details of both stack:
Bluedriod : http://source.android.com/devices/bluetooth.html
Bluez : http://www.kandroid.org/online-pdk/guide/bluetooth.html
3.3.1 We may modify the bluedroid to support our chip.
3.3.2 Working on the bluez support. I'm trying now.
Bluez5 is OK now. See my thread: How to port Bluez to Android 4.4
3.4 Recoveries
With the device tree and kernel, we can build the recovery from source. CWM would be a side product of CM roms and TWRP has a tutorial.
The same View attachment build-k.zip script can be used to build recovery alone. In the rom working direcotry, to build recovery only
type command:
Code:
./build-k p6800 recovery
With no error, recovery.img will be at the same output directory as rom.
Local manifest
https://github.com/danielhk/local_manifest
All model share the same local_manifest.xml
daniel_hk, Check and commit to your tree my patches for PAC-4.2.2
Vaka2 said:
daniel_hk, Check and commit to your tree my patches for PAC-4.2.2
Click to expand...
Click to collapse
Thanks!
It nice to see someone interested in our development!
I have look into your changes briefly.
1. Some changes in media which I'll leave the user to decide. I rather keep the builds as genuie as possible.
2. Some of them are new commit of the original source which will automatically updated in a sync.
3. I forgot to upload a few minor projects to github.
One of which was android_build which will include the files like hciattach building from source. If you study the source of exteranl/bluetooth/bluez, they are there. That means it might have update later. No need to add as a blob.
Actually the best way is including them in PROJECT_PACKAGES section of the device.mk. It would have the same result without touching the original source. Less touch to the original source means fewer conflict later. I was caught up and did't have time to tidy up everything so I leave it there. Also, files from bluetooth should be put in the smdk4210-tab device tree. All model share the same source.
vendor_pac should be upload to official PAC's git. They already gave me an account but they use Google+ which I have trouble to login here. I was working on a job and I hadn't follow up. Now I'm working on 4.4 which PAC haven't release yet. So I put aside first.
You probably didn't have to waste you time if you PM me first.
Everytime I upload to github is a challenge, it always break before complete. I have to monitor it from time to time. Annoying! So I can only upload once a while when everything is ready...
To be honest, I really don't have time to keep up with the old source. The new ports already occupy all my spare times. With limited resource, I can only focus on major issues. Bluetooth after 4.3 is my first priorty. If it can be done, means we can go much further.
If you are interested, you may pick one of the thing I mentioned in this thread to start. Wifi-tethering or the vendor ril library is needed too. You may try it in 4.2.2 first. See what you are good at and interested to. PM me or post again if you have any questions and/or suggestion.
Thanks again for your effort.
daniel_hk said:
Thanks!
It nice to see someone interested in our development!
I have look into your changes briefly.
1. Some changes in media which I'll leave the user to decide. I rather keep the builds as genuie as possible.
2. Some of them are new commit of the original source which will automatically updated in a sync.
3. I forgot to upload a few minor projects to github.
One of which was android_build which will include the files like hciattach building from source. If you study the source of exteranl/bluetooth/bluez, they are there. That means it might have update later. No need to add as a blob.
Actually the best way is including them in PROJECT_PACKAGES section of the device.mk. It would have the same result without touching the original source. Less touch to the original source means fewer conflict later. I was caught up and did't have time to tidy up everything so I leave it there. Also, files from bluetooth should be put in the smdk4210-tab device tree. All model share the same source.
vendor_pac should be upload to official PAC's git. They already gave me an account but they use Google+ which I have trouble to login here. I was working on a job and I hadn't follow up. Now I'm working on 4.4 which PAC haven't release yet. So I put aside first.
You probably didn't have to waste you time if you PM me first.
Everytime I upload to github is a challenge, it always break before complete. I have to monitor it from time to time. Annoying! So I can only upload once a while when everything is ready...
To be honest, I really don't have time to keep up with the old source. The new ports already occupy all my spare times. With limited resource, I can only focus on major issues. Bluetooth after 4.3 is my first priorty. If it can be done, means we can go much further.
If you are interested, you may pick one of the thing I mentioned in this thread to start. Wifi-tethering or the vendor ril library is needed too. You may try it in 4.2.2 first. See what you are good at and interested to. PM me or post again if you have any questions and/or suggestion.
Thanks again for your effort.
Click to expand...
Click to collapse
Thanks to you Daniel for your immense effort on detailing all this! I'm far from a dev status like you and @UpInTheAir. But my support will always go to you developers. Both of you has done an impressive work for an outdated hardware and keep us happy with your Rom and Kernel. Again, thanks a lot.
Thanks, all hail daniel_hk for this!
Daniel,
I can only thank you for all the work he is having to keep our equipment (62xx, 68xx, TXX) alive and really Kitkat will give a new breath, for even in an experimental state already see how the device works smoother. A feather can not help, but I'm looking forward to seeing this in functional status.
As a suggestion, I honestly do not know the procedures, but it would be interesting to ask an official support OmniROM (ROM promising, others liked) and CM. The Galaxy SII has the same hardware, in which the use OmniROM and this magnifies the ROM. Maybe they (Team OmniROM) may offer official support or at least help in these matters!
I'm almost begging for support OmniROM, and Entropy512 (http://forum.xda-developers.com/member.php?u=591147&nocache=1&z=3487892739940434) Kindly reply me: "Someone who is working with one of Those devices and has it working needs to submit it. "
Sorry for my english
Hi Daniel,
Whe I try to repo sync, I was stopped by the error:
"fatal: Couldn't find remote ref refs/heads/cm-10.2"
"Cannot fetch danielhk/proprietary_vendor_samsung_smdk4210."
I was trying to build a cm-10.2 version.
I checked your repository and the folder was right there.
I am new to building android rom. There must be something obvious that I missed.
Thanks.
11nn93n9 said:
Hi Daniel,
Whe I try to repo sync, I was stopped by the error:
"fatal: Couldn't find remote ref refs/heads/cm-10.2"
"Cannot fetch danielhk/proprietary_vendor_samsung_smdk4210."
I was trying to build a cm-10.2 version.
I checked your repository and the folder was right there.
I am new to building android rom. There must be something obvious that I missed.
Thanks.
Click to expand...
Click to collapse
I think you get this line from my PAC manifest.
That was base on cm-10.1, default is cm-10.1. If you take a look of the project danielhk/proprietary_vendor_samsung_smdk4210 on github, you will find only one tree cm-10.1. cm-10.2 can use the same tree. I'll find time to update it later.
When you build cm-10.2, default tree is cm-10.2. The line without a revision will assume the default. That's why you can find it.
Your solution: Add the revision=cm-10.1 at the end of the project.
Code:
.... danielhk/proprietary_vendor_samsung_smdk4210" revision="cm-10.1" />
Good luck!
daniel_hk said:
I think you get this line from my PAC manifest.
That was base on cm-10.1, default is cm-10.1. If you take a look of the project danielhk/proprietary_vendor_samsung_smdk4210 on github, you will find only one tree cm-10.1. cm-10.2 can use the same tree. I'll find time to update it later.
When you build cm-10.2, default tree is cm-10.2. The line without a revision will assume the default. That's why you can find it.
Your solution: Add the revision=cm-10.1 at the end of the project.
Code:
.... danielhk/proprietary_vendor_samsung_smdk4210" revision="cm-10.1" />
Good luck!
Click to expand...
Click to collapse
Thank you for your help. Working on it. Its a big learning curve for me.
I made up to now the most progress with Ubuntu 13.04 64bit version building the cm 10.1 version.
With Ubuntu 12.04 and 13.10, I have been stopped much earlier than on 13.04.
Now I encountered the error below when running build-k p6800:
target thumb C++: libstagefright <= frameworks/av/media/libstagefright/ACodec.cpp
frameworks/av/media/libstagefright/ACodec.cpp:53:24: fatal error: sec_format.h: No such file or directory
compilation terminated.
make: *** [/home/lgeng/android/rom/out/target/product/p6800/obj/SHARED_LIBRARIES/libstagefright_intermediates/ACodec.o] Error 1
make: *** Waiting for unfinished jobs....
I downloaded the folder that contains the sec_format.h file from https://android.googlesource.com/device/samsung/crespo/. But It is not clear to me where to place the folder.
Thanks.
11nn93n9 said:
I made up to now the most progress with Ubuntu 13.04 64bit version building the cm 10.1 version.
With Ubuntu 12.04 and 13.10, I have been stopped much earlier than on 13.04.
Now I encountered the error below when running build-k p6800:
target thumb C++: libstagefright <= frameworks/av/media/libstagefright/ACodec.cpp
frameworks/av/media/libstagefright/ACodec.cpp:53:24: fatal error: sec_format.h: No such file or directory
compilation terminated.
make: *** [/home/lgeng/android/rom/out/target/product/p6800/obj/SHARED_LIBRARIES/libstagefright_intermediates/ACodec.o] Error 1
make: *** Waiting for unfinished jobs....
I downloaded the folder that contains the sec_format.h file from https://android.googlesource.com/device/samsung/crespo/. But It is not clear to me where to place the folder.
Thanks.
Click to expand...
Click to collapse
Not much I can tell. You have to provide the manifest.xml and local_manifest.xml if exist. And device tree changes if exist.
1. build-k is for kernel building only. frameworks_av seems not relevant.
2. crespo is another device. Not the right way to get missing file there.
3. With proper settings and manifest, Ubuntu 12.04 to 13.10 (which I'm using) have no apparent different for building.
If you are building cm-10.1, it is mature. No way there is missing file.
With no other info., I guest there is something wrong with your manifest or device tree. Mixing version projects, wrong flags in device tree, etc.
You might need more reading on what is device tree and manifest first.
Good luck!
Hi,
can you help me with integrating device to manifest.xml? I think there's the problem because everything runs fine until I write luch cm_p6200-userdebug, becuase it shows me that the device isn't found and in /android/rom/device, there isn't any folder called /samsung/p6200. I did everything by the guide, I only changed the local_manifest device info to p6200, because I own it. I'm using Ubuntu 14.04. Thanks for your help.
tom411 said:
Hi,
can you help me with integrating device to manifest.xml? I think there's the problem because everything runs fine until I write luch cm_p6200-userdebug, becuase it shows me that the device isn't found and in /android/rom/device, there isn't any folder called /samsung/p6200. I did everything by the guide, I only changed the local_manifest device info to p6200, because I own it. I'm using Ubuntu 14.04. Thanks for your help.
Click to expand...
Click to collapse
Not much details. next time provide detail. At least, send me your local_manifest.xml. It is definitely the source of your problem. You haven't tell which cm you built too.
I don't know what's wrong because I can't see anything in your side.
But I can tell you that device tree of p6200 didn't download. Check the typings of your local_manifest.xml. It should match the one in github.
Good luck!
TO ANYONE, YOU HAVE TO PROVIDE DETAILS. OTHERWISE, I DON'T KNOW HOW TO HELP!
I really don't have time to guess. You have to forgive me if I ignor your post.
From Nexus 7 上的 Tapatalk
Okay, sorry for it, I didn't know you need it. I want to build a Cm-11. I looked at my local_manifest and I think it's correct. Here is it
edify question for our 815's
Taken from one of your script's
mount("ext4", "EMMC", "/dev/block/mmcblk0p9", "/system");
Installing a .txt into sdcard0/data through .zip in recovery.
I used linux command to find paths for each of the memblocks - but didnt know which path, partition, I am looking for here. /system /userdata etc UHHHH
so to mount to the "sdcard0 /data"directory to install a plain .txt file would be ???
mount("ext4", "EMMC", "??????????????", "/data");
also can I call "/data" anything - like "/datadog" - since this is just a reference to a mount point and is temporary?
One other quicky
First I found this in an edify tutorial -
"/system partition is automatically mounted. this command is needed when you need to mount another partition beside /system or in any condition when recovery mode failed to mount your system partition. the system's location is /dev/stl9, cache /dev/stl10, and data /dev/stl11 (all partition type commonly rfs, will be differ if you have custom kernel or rom)."
since this is a custom rom does /system still auto mount - meaning in ref to to your code above - was it unnecessary but probably good practice or because this is a custom rom there are changes that move the location for /system. I noticed that " the system's location is /dev/stl9, cache /dev/stl10, and data /dev/stl11" was not in the memblock paths??
Start my intro to kernels tomorrow - Oh boy
Off subject - I knew Brazil - tough break with the injury though. Next wins - Argentina and Costa(CoastBaby)!! Going to bed now 6:00 AM WHAT.
Lt.col.johncross said:
Taken from one of your script's
mount("ext4", "EMMC", "/dev/block/mmcblk0p9", "/system");
Installing a .txt into sdcard0/data through .zip in recovery.
I used linux command to find paths for each of the memblocks - but didnt know which path, partition, I am looking for here. /system /userdata etc UHHHH
so to mount to the "sdcard0 /data"directory to install a plain .txt file would be ???
mount("ext4", "EMMC", "??????????????", "/data");
also can I call "/data" anything - like "/datadog" - since this is just a reference to a mount point and is temporary?
One other quicky
First I found this in an edify tutorial -
"/system partition is automatically mounted. this command is needed when you need to mount another partition beside /system or in any condition when recovery mode failed to mount your system partition. the system's location is /dev/stl9, cache /dev/stl10, and data /dev/stl11 (all partition type commonly rfs, will be differ if you have custom kernel or rom)."
since this is a custom rom does /system still auto mount - meaning in ref to to your code above - was it unnecessary but probably good practice or because this is a custom rom there are changes that move the location for /system. I noticed that " the system's location is /dev/stl9, cache /dev/stl10, and data /dev/stl11" was not in the memblock paths??
Start my intro to kernels tomorrow - Oh boy
Off subject - I knew Brazil - tough break with the injury though. Next wins - Argentina and Costa(CoastBaby)!! Going to bed now 6:00 AM WHAT.
Click to expand...
Click to collapse
I don't know what's your objective.
Some points to note:
1. command in updater-script use the function defined in updater-binary. Which are not necessarily a Linux executable command. These commands only valid in the updater-script.
2. You can only mount to an existing directory (mount point) which has not been mounted before. The mount command in Linux is a shell command which is executed by /bin/bash.
3. /system is mount by init following the sequence in .rc files. In recovery, a program is running to provide the interface for I/O. It doesn't require /system partition.
4. No matter which rom (stock or custom), the content of /dev are defined in kernel. Partitions info. is read from the PIT (partition info. table). Kernel will initialize the device at boot accordingly. They are similar to a harddisk partition.
5. our tab has emulated int. sdcard. It is part of the /data, at /data/media.
Hope these might help
earlier question simplified
My problem is with the mount command in a zip file to be used on our Carbon ROM partitions.
My goal is to make 4 different .zip files that can be installed with the recovery software. Each zip places one (the same) simple.txt file into a different folder then the previous zip did.
After I install the 4 zips I should be able to use my file manager app to view the simple.txt in each of the following folders
/sdcard0/data so the ????????? in => mount("ext4", "EMMC", "????????????", "/NAMEOFMOUNT"); would be ==>
/sdcard0/Download so the ????????? in => mount("ext4", "EMMC", "????????????", "/NAMEOFMOUNT"); would be ==>
/Root/data so the ????????? in => mount("ext4", "EMMC", "????????????", "/NAMEOFMOUNT"); would be ==>
/Root/system so the ????????? in => mount("ext4", "EMMC", "????????????", "/NAMEOFMOUNT"); would be ==>
This should give me what i need to fill in what I am missing on understanding this.
Thanks - I hope that this understandable
Off subject - Argentina taken care of biz - Costa is dead to me - Next winners Argentina and Germany
Just about there
6GB with intel I5 -1TB- is now setup as a dual boot with Windows 8.1 and Ubuntu 14.4. All the packages are installed etc.
My linux flavor is ARCH or Manjaro (arch based) - Not a big fan of unity, but since I did not want any pitfalls for support reasons I will use Ubuntu for now.
I now have a descent grasp on GIT and repo, as well as installing cyanogen on one supported device. It would appear that to install on our device any of the custom roms I would just need to provide a local manifest with the proper projects.
I am a little fuzzy on a few things but I think if you could please post two (i815's) local manifest's for two of our newer kitkat roms say Omni & Slim I should be able to figure out my answers or at least narrow down my misunderstandings to exact questions.
Thanks
All Hail GERMANY - - that final game was a real snooze fest.
daniel_hk said:
I don't know what's your objective.
Some points to note:
1. command in updater-script use the function defined in updater-binary. Which are not necessarily a Linux executable command. These commands only valid in the updater-script.
2. You can only mount to an existing directory (mount point) which has not been mounted before. The mount command in Linux is a shell command which is executed by /bin/bash.
3. /system is mount by init following the sequence in .rc files. In recovery, a program is running to provide the interface for I/O. It doesn't require /system partition.
4. No matter which rom (stock or custom), the content of /dev are defined in kernel. Partitions info. is read from the PIT (partition info. table). Kernel will initialize the device at boot accordingly. They are similar to a harddisk partition.
5. our tab has emulated int. sdcard. It is part of the /data, at /data/media.
Hope these might help
Click to expand...
Click to collapse
I'm going to see my mom for 2 days. I'll see if anything missing when return.
Hey guys,
I have been trying to compile the N920C 6.0.1 from the source. Just one time the kernel worked but only when I didn't change anything in menuconfig. If I change even a word in kernel it wont boot up. What am I doing wrong?
My source file : SM-N920C_SEA_MM_Opensource.zip
My firmware file : SM-N920C_N920CXXU2BPB6_CAM_HiepGia_Samsungviet.zip
I was using NDK for toolchain and Linaro. May there be something missing like a tool or may the toolchain be wrong??
"gcc-linaro-4.9-2015.02-3-x86_64_aarch64-linux-gnu.tar.xz" && "android-ndk-r11b-linux-x86_64.zip"
My Makefile setting was like:
Code:
ARCH ?= arm64
CROSS_COMPILE ?= /root/Desktop/ndk/toolchains/aarch64-linux-android-4.9/prebuilt/linux-x86_64/bin/aarch64-linux-android-
After make cmd the output file comes non-compressed called "Image" which is in "arch/arm64/boot/" . Does it make any difference since its not compressed like zImages ?
What is the correct process for converting Image file to boot.img. I have tried both "mkbootimage" and "abootimage" any other way to make boot.img ? Which one is more suitable for my situation?
The process I'm doing is:
Code:
1-)Copy toolchain and source to Kali
2-)Extract kernel source and edit Makefile(for cross_compile)
3-)make ARCH=arm64 ******.defconfig
4-)make menuconfig && save && exit
5-)make ARCH=arm64
6-)Copy "arch/arm64/boot/Image" to another directory
7-)Extract original boot.img from firmware file and then "./split_boot boot.img" with mkbootimg tool
8-)Replace the old kernel file from the boot.img with the new one
9-)Edit /ramdisk/default.prop ====> ro.secure=0 and save
10-)compress ramdisk to gzip
11-) ./mkbootimg --kernel /boot.img-kernel --ramdisk /boot.img-ramdisk.cpio.gz --board SYSMAGIC000KU --base 0x10000000 --pagesize 2048 --ramdiskaddr 0x11000000 -o newboot.img
12-)Tar the newbootimg to flashable zip
13-)Flash it with odin.... Voilaaaa phone won't boot up. Stuck at Samsung Galaxy Note 5 screen:)
What is wrong here?
Please don't tell me to search in forum I have read all of the threads about building kernel. Each one is different from the other so I wanna know what is more suitable in my situation
I know I have asked to many questions but I would appreciate if someone could help me through IRC or some instant messaging service
noone knows ?
Come on guys I have still been waiting for an answer. There are already kernels made by other people in this forum so it must be possible
dewctr said:
Come on guys I have still been waiting for an answer. There are already kernels made by other people in this forum so it must be possible
Click to expand...
Click to collapse
in my opinion it s a problem related to the 'initrd' file..i m diving into chime for now and stucking ; let s converge our efforts to fix all issues with kernel compilation, if u r intressted just pm me.
for info:
initrd provides the capability to load a RAM disk by the boot loader.
This RAM disk can then be mounted as the root file system and programs
can be run from it. Afterwards, a new root file system can be mounted
from a different device. The previous root (from initrd) is then moved
to a directory and can be subsequently unmounted.
initrd is mainly designed to allow system startup to occur in two phases,
where the kernel comes up with a minimum set of compiled-in drivers, and
where additional modules are loaded from initrd.
This document gives a brief overview of the use of initrd. A more detailed
discussion of the boot process can be found in [1].
Operation
---------
When using initrd, the system typically boots as follows:
1) the boot loader loads the kernel and the initial RAM disk
2) the kernel converts initrd into a "normal" RAM disk and
frees the memory used by initrd
3) if the root device is not /dev/ram0, the old (deprecated)
change_root procedure is followed. see the "Obsolete root change
mechanism" section below.
4) root device is mounted. if it is /dev/ram0, the initrd image is
then mounted as root
5) /sbin/init is executed (this can be any valid executable, including
shell scripts; it is run with uid 0 and can do basically everything
init can do).
6) init mounts the "real" root file system
7) init places the root file system at the root directory using the
pivot_root system call
8) init execs the /sbin/init on the new root filesystem, performing
the usual boot sequence
9) the initrd file system is removed
Note that changing the root directory does not involve unmounting it.
It is therefore possible to leave processes running on initrd during that
procedure. Also note that file systems mounted under initrd continue to
be accessible.
---------- Post added at 12:26 PM ---------- Previous post was at 12:18 PM ----------
did u try with latest ubertc toolchain with latest gcc 5.3 compiler?
pm me for info
it is nice to see that someone has also been interested in this thing. Even though Im interested in it, unfortunately, my knowledge is not even a half of yours. I'm not sure how I can be helpful for you but if you ask me things to do, I am ready to apply them. I can try to be helpful with your assistance
Do you want me to try with the toolchain and GCC you mentioned above?
BTW I got it working once without doing any change in defconfig but I don't remember which toolchain I compiled it with. I don't get why it doesn't work when I do changes.