git tips

1. Push to review
    git push ssh://UserName@hostname:29418/repo_name HEAD:refs/for/master(branch name)

2. Push directly, need admin
    git push ssh://UserName@hostname:29418/repo_name HEAD:refs/heads/master

APK tools side note and how to sign a releaseed APK.

###################################
####     How to unpack APK?    ####
###################################

1. Put apktool, apktool.jar under the same folder.
2. Run "apktool d xxx.apk".

############################################
####     How to pack files into APK?    ####
############################################

1. Put apktool, aapt under the same folder.
2. Run "apktool b xxxx-folder-name".
3. The output will be under xxxx-folder-name/dist/ with name xxxx-folder-name.apk.

#######################################
####    How to create keystore?    ####
#######################################

Refer to http://developer.android.com/tools/publishing/app-signing.html
 1. Run "keytool -genkey -v -keystore steven.keystore -alias steven -keyalg RSA -keysize 2048 -validity 10000"
 2. Verity keystore with "keytool -list -v -keystore ./steven.keystore"

###################################
####    How to sign an APK?    ####
###################################

Refer to http://developer.android.com/tools/publishing/app-signing.html
1. Run "jarsigner -verbose -keystore  ./steven.keystore -signedjar xxxx-signed.apk xxxx.apk steven"
   or  "jarsigner -verbose -sigalg SHA1withRSA -digestalg SHA1 -keystore steven.keystore ./xxxx.apk"
2. Verify the signed APK with "jarsigner -verify -verbose ./xxxx.apk"

###################################
####    How to laing an APK?   ####
###################################

Refer to http://developer.android.com/tools/publishing/app-signing.html
1. Run "zipalign -v 4 xxx-unaligned.apk xxx.apk"

zipalign ensures that all uncompressed data starts with a particular byte alignment
relative to the start of the file, which reduces the amount of RAM consumed by an app.

NDK build specific module

/PATH/TO/ndk-build APP_MODULES='name'

git & gerrit setup in Ubuntu

For gerrit install,
1. Download gerrit from http://gerrit-releases.storage.googleapis.com/index.html
    I choose gerrit-2.11.2.war
2. Create mySQL for gerrit
    a. mysql -u root -p PASSWD
    b.Create a Gerrit specific user within the database and assign it a password,  create a database, and give the user full rights:

"CREATE USER 'gerrit2'@'localhost' IDENTIFIED BY 'secret';"
"CREATE DATABASE reviewdb;"
"ALTER DATABASE reviewdb charset=latin1;"
"GRANT ALL ON reviewdb.* TO 'gerrit2'@'localhost';"
"FLUSH PRIVILEGES;"

3. Install gerrit
 java -jar gerrit-2.11.2.war init -d /GERRIT/INSTALL/PATH

4. Choose HTTP as authentication method and set rest as default
5. Install apache2
6. Create a virtual host with the following httpd.conf under  /etc/apache2/sites-available and create a symlink to /etc/apache2/sites-enabled

LoadModule proxy_module /usr/lib/apache2/modules/mod_proxy.so
LoadModule proxy_http_module /usr/lib/apache2/modules/mod_proxy_http.so

<VirtualHost *:8082>
  ServerName localhost
  ProxyRequests Off
  ProxyVia Off
  ProxyPreserveHost On
  DocumentRoot /var/www/html

  <Proxy *>
    Order deny,allow
    Allow from all
  </Proxy>

  <Location /login/>
    AuthType Basic
    AuthName "Gerrit Code Review"
    AuthBasicProvider file
    AuthUserFile /home/gerrit2/review_gerrit/etc/gerrit.htpasswd
    Require valid-user
    Order Deny,Allow
    Allow from all
  </Location>

  AllowEncodedSlashes On
  ProxyPass / http://SITESURL:6267/ nocanon
</VirtualHost>

 7. Create HTTP basic auth user/passwd file, later you won't need -c flag.(-c: create)

sudo htpasswd -c /home/gerrit2/review_gerrit/etc/gerrit.htpasswd UserName

8. My final gerrit.config is as following

[gerrit]
    basePath = git
    canonicalWebUrl = http://SITESURL:6267
[database]
    type = mysql
    hostname = localhost
    database = reviewdb
    username = gerrit2
[index]
    type = LUCENE
[auth]
    type = HTTP
    loginUrl = http://SITESURL:8082/login/
[sendemail]
    smtpServer = localhost
[container]
    user = gerrit2
    javaHome = /usr/lib/jvm/java-7-openjdk-amd64/jre
[sshd]
    listenAddress = *:29418
[httpd]
    listenUrl = http://SITESURL:6267/
[cache]
    directory = cache
[plugins]
    allowRemoteAdmin = true

9. Link to http://SITESURL:6267 and it will pop up a dialogue, fill in the username and passwd just create with htpasswd and change the settings in gerrit webUI. For example, paste your ssh public key and change your full name

kernel scheduler entry point

1. Upon returning to user-space or returning from an interrupt, the need_resched flag is checked. If it is set, the kernel invokes the scheduler before continuing.
2. Process preemption
3. Running out of time slice
4. Waiting for event
5. Waiting for IO
6. In idle loop cpu_idle_loop() (from code sched/idle.c)

In short, user preemption can occur
1. When returning to user-space from a system call
2. When returning to user-space from an interrupt handler

Kernel preemption:
So when is it safe to reschedule? The kernel can preempt a task running in the kernel
so long as it does not hold a lock.That is, locks are used as markers of regions of nonpre-
emptibility. Because the kernel is SMP-safe, if a lock is not held, the current code is reen-
trant and capable of being preempted.
The first change in supporting kernel preemption was the addition of a preemption
counter, preempt_count , to each process’s thread_info .This counter begins at zero and
increments once for each lock that is acquired and decrements once for each lock that is
released.When the counter is zero, the kernel is preemptible. Upon return from interrupt,
if returning to kernel-space, the kernel checks the values of need_resched and
preempt_count . If need_resched is set and preempt_count is zero, then a more impor-
tant task is runnable, and it is safe to preempt.Thus, the scheduler is invoked. If
preempt_count is nonzero, a lock is held, and it is unsafe to reschedule. In that case, the
interrupt returns as usual to the currently executing task.When all the locks that the cur-
rent task is holding are released, preempt_count returns to zero.At that time, the unlock
code checks whether need_resched is set. If so, the scheduler is invoked.

In short, Kernel preemption can occur

1. When an interrupt handler exits, before returning to kernel-space
2. When kernel code becomes preemptible again
3. If a task in the kernel explicitly calls schedule()
4. If a task in the kernel blocks (which results in a call to schedule() )


Unlike softirqs, however, two of the same tasklets never run concurrently—although two different tasklets can run at the same time on two different processors


Recall that two tasklets of the same type do not ever run simultaneously.Thus, there is no need to protect data used only within a single type of tasklet. If the data is shared between two different tasklets, however, you must obtain a normal spin lock before accessing the data in the bottom half.You do not need to disable bottom halves because a tasklet never preempts another running tasklet on the same processor.
With softirqs, regardless of whether it is the same softirq type, if data is shared bysoftirqs, it must be protected with a lock. Recall that softirqs, even two of the same type, might run simultaneously on multiple processors in the system. A softirq never preempts another softirq running on the same processor, however, so disabling bottom halves is not needed.

Install SCIM chewing on Ubuntu 14.04

sudo apt-get install scim scim-chewing
sudo im-config -s scim-bridge

Update/Recover from Ubuntu 14.04 upgrade

1. First of all, the upgrade is simple. Follow the prompt and click next.

2. During the process, it says my xserver-xcore-video-all package has problem and my upgrade is canceled. In order to go on, I deleted that package

sudo apt-get remove xserver-xorg-video-all

3. Things go well until the system reboots

4. Download the latest NVIDIA driver from http://www.nvidia.com/Download/index.aspx and install.
    p.s. The latest one is 340.24 and before this I tried the previously used the old one 331.20 and a random one not even double checked 331.67, both won't work and says "The system is running in low-graphis mode."
    p.s. Please check /var/log/Xorg.0.log and kern.log for detail description. (words regarding nvidia driver problems)

5. Now we can log in but the resolution is very weired and shows only desktop. No unity icon, right botton does not work.
    p.s. I thought it's unity/compiz/lightdm problem so I removed the previous settings by

rm -rf  /.config/*
rm -rf ~/.compiz/*
rm -rf ~/.cache/compizconfig-1/  

but I think the above is unrelated.

6. Step 5 does not solve my problem and there is still not desktop with unity and I checked /var/log/lightdm/lightdm.log and found "Xlib:  extension "GLX" missing on display ":0"."

7. "DISPLAY=:0 glxinfo" shows the same error

8. ldd /usr/bin/glxinfo and found libGL.so links to correct version(340.23) driver gl library.

9. Suddently I recalled that I removed xservice-xorg-video-all in the first place.

10. Trying to install this package and found the PPAs are different for 13.10 and 14.04. So I remove the the original ppa

sudo ppa-purge xorg-edgers

and add it back

sudo apt-add-repository ppa:xorg-edgers

sudo apt-get update

sudo apt-get upgrade 

11. Finally my Ubuntu upgrade finishes.

SOP to reinstall nvidia graphic card driver

From time to time, the display won't work correctly after my system upgrade or software upgrade. The following is the SOP to solve this problem:
a.
   1. don't use Nvidia's graphic card

b.
   1. ctrl + alt + F1 to change to another tty
   2. sudo service lightdm stop to stop X
   3. sudo ./XXXX-nvidia.xxx.run
   4. sudo service lightdm start to activate X

LKD notes

1. types of exceptions:  divided by 0; page fault; system call
2. Why interrupt can not sleep?  Interrupt do not have task_info structure and therefore can not trace through the rbtree to find the next runnable process to run. Though current macro points to the interrupted process, but there is no way to find the interrupt handler that sleeps and to continue execution.
3. When do_IRQ() finished the isr, the entry code will then call ret_from_intr() and it checks is a reshcedule is pending(need_resched is set). If the reschedule is pending and the kernel is returning to user-space(that is, the interrupt interrupted a user process), schedule() is called. If the kernel is returning to kernel-space(that is, the interrupt interrupted the kernel itself), schedule() is called only if the preempt_count is zero. After schedule() returns, or if there is no work pending, the initial registers are restored and the kernel resumes whatever was interrupted. (do_IRQ() corresponds to handle_IRQ() under arch/arm/kernel/irq.c for ARM)
4. Sometimes we see spin_lock_irqsave()/spin_unlock_irqrestore() and mutex_lock()/mutex_unlock() pairs and the first is to provide protection against concurrent access from a possible interrupt handler and the later is to provide protection against concurrent access from another processor.
5. Interrupt handlers run asynchronously with at least the current interrupt line disabled.

Monitor dd progress

When we flash an image to a SD card through dd command,

we won't be able to know the progress by default.

If you need type sudo kill -USR1 $(pgrep ^dd) 

and you will see the progress.

Steps to mount .img

Ok, you may have a image with multi-partitions(for example, boot and rootfs )and you may need to access the data under rootfs, and, you don't want to flash this image.

The concept is you mount the partition inside the image you want and the following are steps.

1. find partition information

fdisk -l xxx.img

2. mount the partition as block device and have to specify the start address in bytes

 sudo losetup /dev/loop0 xxx.img -o $((Start Sectors*512))

3. mount it

sudo mount -t ext3 /dev/loop0 /mount/point

mount -o loop /path/to/image won't work in this case since the start byte does not contain file system information in this image type

Build OpenNI apk in Ubuntu shell

You should have AndroidManifest.xml in the APK root first and use the following command to generate build.xml

~/ADT/android-sdk-linux/tools/android update project --target android-18 -p .

Now you can build

 ant debug/release -Dsdk.dir=/home/stevenchiu/ADT/android-sdk-linux/

Easy~

To run OpenNI Simple Viewer wth Kinect in BeagleBone Black Android

 1. Build OpenNI

git clone git://github.com/OpenNI/OpenNI

git checkout -b 1.5.4.0 Unstable-1.5.4.0

 checkout to tag 1.5.4.0, only this tag works for me

cd Platform/Android/jni

${NDK_ROOT}/ndk-build

 You have to define your NDK_ROOT of course. I use NDK r7

 If you encountered cpu_set_t can not be recognized in  NiSkeletonBenchmark.cpp, simply ignore this behaviour

 -#if (XN_PLATFORM != XN_PLATFORM_MACOSX)       
+#if (XN_PLATFORM != XN_PLATFORM_MACOSX && XN_PLATFORM != XN_PLATFORM_ANDROID_ARM)      ^M
     //we want out benchmark application will be running only on one CPU core
     cpu_set_t mask;



 2. Build SensorKinect 

export NDK_MODULE_PATH=/path/to/openni/Platform/Android/jni

git clone git://github.com/avin2/SensorKinect.git

git checkout -b unstable origin/unstable

cd Platform/Android/jni

${NDK_ROOT}/ndk-build

 3. Upload your libraries and binary to Android
    Now we have to upload the libraries located in Platform/Android/libs/armeabi-v7a/*.so to android /system/libs. We have to upload OpenNI libraries first and then SensorKinect since some libraries should be replaced according to the HW.


i.e. Kinect in this case.
    
    Please remember to upload the Sample-SimpleRead binary to /system/bin/



 4. Mount USBFS to grant USB device permission
   mount -o devmode=0666 -t usbfs none /proc/bus/usb


 5. Push xmls to /Data/ni
     adb push SamplesConfig.xml /Data/ni
      adb push modules /Data/ni
      adb push GlobalDefaultskinect.ini /Data/ni
      adb push License.xml /Data/ni

6. Run Sample-Read under /Data/ni
    You should be able to see the middle point and frame rate

#Trouble shooting
    USB open fail => sudo rmmod  gpsca_kinect

To run Nite samples in Ubuntu 13.10 (Nite + OpenNI + SensorKinect)

1. Build OpenNI

git clone git://github.com/OpenNI/OpenNI

git checkout -b 1.5.4.0 Unstable-1.5.4.0

 checkout to tag 1.5.4.0, only this tag works for me

cd Platform/Linux/CreateRedist

./RedistMaker

cd ../Redist/$(Openni Version)/

In my case, it's OpenNI-Bin-Dev-Linux-x64-v1.5.4.0

sudo install.sh

2. Build SensorKinect

git clone git://github.com/avin2/SensorKinect.git

git checkout -b unstable origin/unstable

cd Platform/Linux/CreateRedist

./RedistMaker

 cd ../Redist/$(Sensor Kinect Version)/

In my case, it's  Sensor-Bin-Linux-x64-v5.1.2.1

sudo install.sh

It's that simple.

now you can download NITE-Bin-Dev-Linux-x64-v1.5.2.21.tar.zip and run samples under Samples/Bin/x64-Release

[BBB] memo of configuring BeagleBone Black network

manually add a gateway

route add default gw 192.168.1.1 dev eth0 => add default gateway 
 

The following configuration in /etc/network/interfaces does all above:

#iface eth0 inet dhcp
iface eth0 inet static
address 192.168.1.2
netmask 255.255.255.0
network 192.168.1.0
broadcast 192.168.1.255
gateway 192.168.1.1
dns-nameservers 192.168.1.1 

Finally, add nameserver in /etc/resolf.conf

echo "nameserver 192.168.1.1" > /etc/resolv.conf 

Restart network now:
/etc/init.d/networking restart =>  restart network or 

ifdown eth0; ifup eth0

toolchain for BBB is arm-linux-gnueabihf

Problem solved! 

extract and cross compiled with CC=arm-linux-gnueabihf-gcc ./configue --target=arm-linux 

Build OpenCV from souce

sudo apt-get install libgtk2.0-dev

I only encounter this problem. please google the error message you see to get the corresponding solutions

git clone https://github.com/Itseez/opencv.git

cd opencv

git checkout 2.4.6.2 -b 2.4.6.2

default is master branch and I manually checkout to 2.4.6.2 tag.
mkdir build(whatever name you want)

cd build

cmake-gui ..

(to generate the configurations you want and generate make file. sudo apt-get install cmake-gui if you don't have it.)

make; sudo mak install

Generate patches for specific user

To generate my own patches by the following command:

for commit in `git log --author=XXX --pretty=oneline | awk '{print $1}'`; do git format-patch -1 "$commit"; done

command line access windows share

無法查看此摘要。請 按這裡查看文章。

some string related tips

0. replace strings for all files in a folder
find ./ | xargs grep -l "$STRING_TO_BE_REPLACED" | xargs sed -i -e "s/$STRING_TO_BE_REPLACED/$STRING_TO_REPLACE/g"

0. find files and then grep som key word
find ./ -iname '$FILE_NAMES*' | xargs grep $SEARCH_KEY_WORDS

0. find files and then remove
find ./ -iname '$FILE_NAMES*' | xargs rm -rf

0. find and push multiple files from adb
find ./ -iname *.ko  | xargs -t -i adb push {} /system/lib/modules/

0. Search multiple words in VIM
/\(kernel\|panic\);  you will search both kernel and panic at the same time

0. Copy and rename multiple filse
for i in `find ./ -iname "*file_name*"`; do cp $i `echo $i | sed "s/$STRING_TO_BE_REPLACED/$STRING_TO_REPLACE/g"`; done

Note about kobjects, ksets, and ktypes.

kobject, quoted form kernek documentation: (example in samples/kobject/kobject-example.c)
1. A kobject must be initialized. for example

void kobject_init(struct kobject *kobj, struct kobj_type *ktype);
The ktype is required for a kobject to be created properly, as every kobject
must have an associated kobj_type.

2. After calling kobject_init(), to regsiter the kobject with sysfs, kobject_add() must be called
int kobject_add(struct kobject *kobj, struct kobject *parent, const char *fmt, ...);
This setup the parrent of the kobject and the name for the kobject properly. If the kobject is to be associated with a specific kset, kobj->kset must be assigned before calling kobject_add(). If a kset is associated with a kobject, then the parent for the kobject can be set to NULL in the call to kobject_add() andthen the kobject's parent will be the kset itself.
or call int kobject_init_and_add(struct kobject *kobj, struct kobj_type *ktype,
struct kobject *parent, const char *fmt, ...); to init and add kobject to the kernel at the same time.

3. To inform userspace that a kobj has been created, call
int kobject_uevent(struct kobject *kobj, enum kobject_action action);

4. To create a simple directory in the sysfs hierarchy and not to mess with the whole complication of ksets, show and store functions, call
struct kobject *kobject_create_and_add(char *name, struct kobject *parent);
It creates a kobject and place it in sysfs in the location underneath the specified parent kobject.

5. To release a kobject, don't do kfree() but kobject_put(). A good practice is to use kobject_put() as an error check after kobject_init() to avoid errors creeping in.

6. Every kobject must have a release() method and the release() method is not stored in the kobject itself but associated with the ktype. for example:
struct kobj_type {
void (*release)(struct kobject *);
const struct sysfs_ops *sysfs_ops;
struct attribute **default_attrs;
};
This structure must be referenced when you call kobject_init() or kobject_init_and_add()..