Vim stuff

TAB completion

The following lines should be added to .vimrc and .gvimrc to make vim's auto-completion work just like bash:

set wildmode=longest,list

set wildmenu

Text width

Vim does not break text into lines automatically. This is good for programming, but not so much for documents. To make vim break the text into lines set the textwidth property:

set textwidth=70

When you do that, vim will apply the property to the current text. If you want to set the property for the entire document go to the top, enter gq, and then go to the bottom:

gg

gq

G

The gq command applies the format to a text segment marked by 'movement'.

Working with tabs

Edit file in a new tab:

:tabe

Next and previous tab:

gt

gT

Bookmarks:

Mark position

: m {a-z}

Return to marked position

: ' {a-z}

Return to previous position

: ' '

Editing remote files:
tabe scp://hostname//path/to/file
or
vim scp://hostname//path/to/file

Patching

Patching code is really easy. Let's look at some examples. Consider the patch file parsec-ocreat.patch, which has the following beginning lines:

diff -N -u -r parsec-2.1-orig/pkgs/kernels/dedup/src/decoder.c parsec-2.1/pkgs/k

ernels/dedup/src/decoder.c

--- parsec-2.1-orig/pkgs/kernels/dedup/src/decoder.c 2009-01-27 17:09:55.0000

00000 -0800

+++ parsec-2.1/pkgs/kernels/dedup/src/decoder.c 2009-09-01 00:51:12.000000000 -0

700

The first file --- parsec-2.1-orig/pkgs/kernels/dedup/src/decoder.c represents the original file which will be patched. The second file +++ parsec-2.1/pkgs/kernels/dedup/src/decoder.c is the one with the correct modifications. We won't care about the second file, only the first one. To patch the original file we need to match the location given in the "---" line of the patch file with its actual location in our directory tree. In this case I don't have the directory parsec-2.1.orig, but I do have parsec-2.1. To make them match I do the following.

$ cd parsec-2.1

$ patch -p1 <parsec-ocreat.patch

The p# parameter defines the number of slashes you have to remove from the path to make the two files match. In this case, because the first directory doesn't match I used p1. Put another way:

-p1 "parsec-2.1-orig/pkgs/kernels/dedup/src/decoder.c" = pkgs/kernels/dedup/src/decoder.c

If you're afraid of patching the wrong file or are unsure of how patch works, do a dry run first:

$ patch --dry-run -p1 <parsec-ocreat.patch

If you receive successful messages you can now patch your file. Let's look at another example, the file blackscholes.c.patch which begins as:

--- blackscholes.c 2009-07-28 00:58:12.000000000 +0100

+++ blackscholes.new.c 2010-05-04 12:14:07.416033100 +0100

This time the path is not given, so I need to move my prompt all the way where the blackscholes.c file is. Now the name of the original file will match the name given in the "---" line of the patch file. As a result, no slashes need to be removed:

$ patch -p0 <blackscholes.c.patch

Reversing a patch is also possible. Just run patch with the -R option:

$ patch -R -p0 <blackscholes.c.patch

This will restore the previous original file.

That's all. Good luck!

References:

http://www.linuxtutorialblog.com/post/introduction-using-diff-and-patch-tutorial

http://linux.byexamples.com/archives/163/how-to-create-patch-file-using-patch-and-diff/

SSH stuff

SSH key

This is how you can set an ssh key so that password will not be required.

From the client enter:

$ ssh-keygen -t dsa

This will generate the public key, which is saved in the file ~/.ssh/id_dsa.pub, and the private key, saved in ~/.ssh/id_dsa. Now we copy the public key in the server by entering:

$ ssh-copy-id username@remotehost

If you use a non-standard port a little trick is necessary. We put the host address in quotation marks as

$ ssh-copy-id '-p 2222 username@remotehost'

That should work just fine! A good reference is https://help.ubuntu.com/9.04/serverguide/C/openssh-server.html and http://mikegerwitz.com/2009/10/07/ssh-copy-id-and-sshd-port/

Screen

This is how you use screen so that you can keep jobs running at a remote host even when the connection drops or you logout. First install screen on the server:

$ sudo apt-get install screen

When you log in via ssh you can initiate a screen session by typing $ screen. Now you can play around with screen using the following commands:

Ctrl+a c : creates a new terminal tab within the screen session

Ctrl+a p : goes to the previous terminal tab

Ctrl+a n : goes to the next terminal tab

Ctrl+a d : detaches the current screen so that you can come back to it later

exit : exits a terminal tab within screen or exits the screen if there is only one tab open

Within a screen you can create several terminals to interact with the host system. This is done with the command Ctrl+ c. You can also multiple screen running in the same system.

To list all screens running in the system use:

$ screen -ls

The output will be something like this:

There are screens on:

13528.pts-1.brazil (09/14/2010 09:22:08 PM) (Attached)

13496.pts-1.brazil (09/14/2010 09:09:02 PM) (Detached)

2 Sockets in /var/run/screen/S-gbezerra.

To reconnect to the detached screen type screen -r :

$ screen -r 13496.pts-1.brazil

When a screen is detached you can logout and it will still be running. The same occurs if the connection drops. Just log back in, list the active screens and reconnect to the screen you want.

You can also name the screen you create, so that it is easier to reconnect to it.

$ screen -S george

To reconnect enter:

$ screen -r george

Finally if you're running a screen in a given terminal/client and want to capture that screen at another terminal/client, but the screen has not been detatched, enter:

$ screen -rd george

this will detatch the screen and pull it into your current terminal.

A good reference for screen is http://www.howtoforge.com/linux_screen

ssh stuff

Connecting directly to my brazil machine:

$ ssh -p 3335 adaptive....

Compiling the Kernel

This are the steps I followed to compile the linux kernel 2.6.35.4 on Ubuntu 10.04.

1) Download. The first step is to download the kernel at http://www.kernel.org/ and then unpack it in some location.

2) Prerequisites. Now we make sure all the necessary packets are installed:

$ sudo apt-get install fakeroot kernel-wedge build-essential makedumpfile kernel-package

$ sudo apt-get install libncurses5 libncurses5-dev

3) Configuration. Building the kernel is all about configuration. Configuring the kernel for your machine is fun but takes a lot of time and might be dangerous if you don't know what you're doing. We will take a shortcut here and use the Ubuntu configuration file for the current kernel. Go to the source directory and follow the steps bellow:

$ cd linux-2.6.35.4

$ cp /boot/config-2.6.32-24-generic .config

$ make menuconfig

In the configuration menu you can see all the options for your kernel. I only changed the processor from generic to "Core 2/newer Xeon" and the preemption model to make my computer more responsive. Fell free to explore other options. The help feature can be very useful to learn more about your computer.

4) Initramfs and initrd. Now we need a couple more tweaks before we start. First we will borrow some scripts that will help us create the initramfs file system.

$ sudo cp /usr/share/doc/kernel-package/examples/etc/kernel/postinst.d/initramfs /etc/kernel/postinst.d/initramfs

$ sudo cp /usr/share/doc/kernel-package/examples/etc/kernel/postrm.d/initramfs /etc/kernel/postrm.d/initramfs

Next we need to make sure that the initrd image is also created. To do so follow the steps below:

First create your overlay directory:

$ cp -r /usr/share/kernel-package $HOME

Now download the source code for your current kernel version:

$ sudo apt-get source linux-image-2.6.32-24-generic

This will unpack the source in $HOME/linux-2.6.32. Now copy the control scripts into your new overlay:

$ cp linux-2.6.32/debian/control-scripts/{postinst,postrm,preinst,prerm} kernel-package/pkg/image/

$ cp linux-2.6.32/debian/control-scripts/headers-postinst kernel-package/pkg/headers/

5) Parallelizing. Compiling the kernel might take a long time, like 2 hours. So if you have multiple processors in your machine you should take advantage of them to speedup the process. To do so, setup the concurrency level variable to the number of processors you have plus 1. In my case I have a core i3, with 2 processors but also with hyperthreading, so my computer sees 4 processors.

$ export CONCURRENCY_LEVEL=5

6) Compilation. Now everything should be ready for the grand compilation. Type the commands below to start compiling:

$ make-kpkg clean

$ fakeroot make-kpkg --initrd --overlay-dir=$HOME/kernel-package --append-to-version=-gbezerra kernel-image kernel-headers

Change the "gbezerra" string to whatever you want to be appended to your kernel name.

7) Installation. After 30min, depending on how many processors you have, you should have your kernel compiled and some .deb packages ready to be installed. Go to one directory above and proceed as follows.

$ cd..

$ sudo dpkg -i linux-image-2.6.35.4-gb_2010.09.10_amd64.deb

$ sudo dpkg -i linux-headers-2.6.35.4-gb_2010.09.10_amd64.deb

After that your kernel should be installed. Just to make sure, go to the /boot directory and see that all the new kernel files are there. You're ready to reboot your machine and try it out.

$ sudo reboot

Good luck!

For a helpful reference go to https://help.ubuntu.com/community/Kernel/Compile

Updating database

In order for the 'locate' command to work we need to update the database by simply typing 'updatedb'

Adding user to sudoers

Simple stuff I don't wanna forget:

sudo adduser username admin

Or go to /etc/sudoers and add:

user ALL=(ALL) ALL